Sender: azz@cartman.azz.net
Newsgroups: alt.folklore.computers
Subject: Features to save from old OSs
Mime-Version: 1.0 (generated by tm-edit 7.106)
Content-Type: text/plain; charset=US-ASCII
From: Adam Sampson <azz@gnu.org>
Message-ID: <873dlsdrci.fsf@cartman.azz.net>
Lines: 21
X-Newsreader: Gnus v5.5/Emacs 20.2
Date: 03 Jul 2000 00:50:37 +0100
NNTP-Posting-Host: 212.159.22.152
X-Complaints-To: abuse@plus.net.uk
X-Trace: stones 962581772 212.159.22.152 (Mon, 03 Jul 2000 00:49:32 BST)
NNTP-Posting-Date: Mon, 03 Jul 2000 00:49:32 BST
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!News.Amsterdam.UnisourceCS!skynet.be!newsfeed.skycache.com!Cidera!63.211.125.72!cyclone2.usenetserver.com!news-out.usenetserver.com!btnet-peer0!btnet!landlord!stones.POSTED!cartman.azz.net!nobody
Xref: chonsp.franklin.ch alt.folklore.computers:59017

Hiya.

For an increasing number of people --- me included --- the only
non-microcomputer-originated OS that they've ever used is Unix. While
Unix is a good place to start, there are plenty of interesting and
useful features around in older or less popular systems which don't
get considering when working on free OSs.

If you were building a new free OS, what features would you like from
other (non-Unix) OSs? I'd be very interested to hear the answers; this
follows a similar discussion on the debian-hurd mailing list a while
back. Features that came up: block-orientated IO for database
operations, pervasive message queueing (as in QNX and AmigaOS),
filesystem versioning, and having debugging features built into the
shell. However, the breadth of experience here is much wider --- let
me know what you'd like to see...

-- 

Adam Sampson
azz@gnu.org

######

From: jmfbahciv@aol.com
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: Mon, 03 Jul 00 08:07:22 GMT
Organization: UltraNet Communications, Inc.
Lines: 27
Message-ID: <8jprr5$9ds$1@bob.news.rcn.net>
References: <873dlsdrci.fsf@cartman.azz.net> <877lb4f18g.fsf@litterbox.meowing.net>
X-Trace: kFX8xly14t4lGjYk++jWgC1inzNtmxeEZgWCiORD30Q=
X-Complaints-To: abuse@rcn.com
NNTP-Posting-Date: 3 Jul 2000 11:01:57 GMT
X-Newsreader:  News Xpress Version 1.0 Beta #4
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news-fra1.dfn.de!news0.de.colt.net!colt.net!diablo.theplanet.net!europa.netcrusader.net!207.172.3.37!feed1.news.rcn.net!rcn!207-172-255-136
Xref: chonsp.franklin.ch alt.folklore.computers:59001

In article <877lb4f18g.fsf@litterbox.meowing.net>,
   Extra Strength Flufferin <supersede@india.com> wrote:
>Adam Sampson <azz@gnu.org> wrote:
>
>> If you were building a new free OS, what features would you like from
>> other (non-Unix) OSs?
<snip>

>a proper usage accounting system

Heh, I know how to do that.

Just about everything that I snipped was already done 
"properly".

The question posed here isn't really a good question.  No
operating system can be all things to all people.  I think
the type of installation may be defined by the type of
[may I call it] data that needs to be processed.  And I don't
mean data as in database applications.  

/BAH


/BAH

Subtract a hundred and four for e-mail.

######

From: Extra Strength Flufferin <supersede@india.com>
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: 02 Jul 2000 21:31:43 -0400
Organization: Easier on your stomach than ground glass
Lines: 25
Message-ID: <877lb4f18g.fsf@litterbox.meowing.net>
References: <873dlsdrci.fsf@cartman.azz.net>
X-Complaints-To: newsabuse@supernews.com
Mail-Copies-To: never
X-Meow: Wouf
X-Face: #!n`o'kqzEiF(\CL4IR$H.F-||!S*5wv~g8~prn}Z<+F/-^?oJ:#_V#QP?G}!!yayV0]i\'   GTb6:]Tic*!vjH}wm>ILhZBn8U0XEjh'l~yrjB?4Iiph5N9:kwqd/}86UTr9i|LCu]Bd)~G1R9U(qJ   KOyEQK8*?
User-Agent: Gnus/5.0807 (Gnus v5.8.7) XEmacs/21.1 (Capitol Reef)
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!news.datacomm.ch!newsmaster-01.atnet.at!atnet.at!newsrouter.chello.at!newsfeed01.sul.t-online.de!newsfeed00.sul.t-online.de!t-online.de!diablo.theplanet.net!news.maxwell.syr.edu!newsfeed.stanford.edu!sn-xit-03!supernews.com!sn-inject-01!corp.supernews.com!not-for-mail
Xref: chonsp.franklin.ch alt.folklore.computers:59042

Adam Sampson <azz@gnu.org> wrote:

> If you were building a new free OS, what features would you like from
> other (non-Unix) OSs?

reliable signal handling

more attention to data integrity, particularly in the disk department
(the versioning you mention is a step in the right direction)

more attention to keeping resource overconsumption in check (AIX
actually gets some of this one right)

a sane interface to networks

clustering that actually works

a dbm that doesn't suck

a modern documentation system to replace those awkward man pages

a proper usage accounting system

a more flexible security layer (setuid ends up being the answer to too
many problems and it's too easy to get wrong)

######

From: "George R. Gonzalez" <grg1@flash.net>
Newsgroups: alt.folklore.computers
References: <873dlsdrci.fsf@cartman.azz.net> <877lb4f18g.fsf@litterbox.meowing.net>
Subject: Re: Features to save from old OSs
Lines: 45
X-Newsreader: Microsoft Outlook Express 4.72.3110.1
X-MimeOLE: Produced By Microsoft MimeOLE V4.72.3110.3
Message-ID: <cNT75.4138$7%3.326535@news.flash.net>
Date: Mon, 03 Jul 2000 03:30:16 GMT
NNTP-Posting-Host: 38.31.46.118
X-Complaints-To: abuse@flash.net
X-Trace: news.flash.net 962595016 38.31.46.118 (Sun, 02 Jul 2000 22:30:16 CDT)
NNTP-Posting-Date: Sun, 02 Jul 2000 22:30:16 CDT
Organization: FlashNet Communications, http://www.flash.net
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news.maxwell.syr.edu!nntp.flash.net!news.flash.net!not-for-mail
Xref: chonsp.franklin.ch alt.folklore.computers:59010

Some of the nice features from CDC's KRONOS and NOS operating systems.
That's not to say that some of these were not double-edged swords.

*  Local files.   Each task's files were in their own namespace, and went
away
when the job ended, unless you explicitly attached a global file, or saved
the local
file to permanent storage (EXTREMELY double-edged).  But it meant much
less muss and fuss cleaning up after a job was run.

* job control statements separate from input data stream.

*  "MODIFY", a simple and very powerful way of maintaining source code,
with each revision identified and undo-able at will.

* ACL lists for all permanent files.

* Access logs for files.

* "Dayfiles",a nice summary of job steps executed.

* Fast I/O.  You could fill up a roomful of disks in just a few seconds!

* concurrent system requests.  You could ask that any system request either
wait for completion, or return to you right away.   (Later trimmed back to
just
I/O requests)

* 60 bits of user access rights.

* terminal monitoring, logoff ability.

* Good set of system macros.


----

There were quite a few misdesigns too, but let's not dwell on those...

######

From: "Bob Billing (AKA Uncle Bob)" <unclebob@tnglwood.demon.co.uk>
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: Mon, 03 Jul 2000 08:49:42 +0100
Message-ID: <39604596.3ECE7EC0@tnglwood.demon.co.uk>
References: <873dlsdrci.fsf@cartman.azz.net>
NNTP-Posting-Host: tnglwood.demon.co.uk
X-NNTP-Posting-Host: tnglwood.demon.co.uk:158.152.132.30
X-Trace: news.demon.co.uk 962610995 nnrp-06:18761 NO-IDENT tnglwood.demon.co.uk:158.152.132.30
X-Complaints-To: abuse@demon.net
X-Mailer: Mozilla 4.72 [en] (X11; U; Linux 2.2.14-5.0 i586)
X-Accept-Language: en
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Lines: 13
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news-fra1.dfn.de!news0.de.colt.net!colt.net!dispose.news.demon.net!demon!news.demon.co.uk!demon!tnglwood.demon.co.uk!falstaff.tanglewood!nobody
Xref: chonsp.franklin.ch alt.folklore.computers:59026

Adam Sampson wrote:

> If you were building a new free OS, what features would you like from
> other (non-Unix) OSs? I'd be very interested to hear the answers; this

I'd like to have a DCL clone shell. I found shell programming far easier
on the VAX than on unix.

-- 
I am Robert Billing, Christian, inventor, traveller, cook and animal
lover, I live near 0:46W 51:22N.  http://www.tnglwood.demon.co.uk/
"It burned me from within. It quickened; I was with book as a woman
is with child." CS Lewis - Till we have faces, Ch 21.

######

Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
References: <873dlsdrci.fsf@cartman.azz.net> <39604596.3ECE7EC0@tnglwood.demon.co.uk>
Organization: None
X-Newsreader: trn 4.0-test72 (19 April 1999)
From: kragen@dnaco.net (Kragen Sitaker)
Lines: 18
Message-ID: <8_X75.24474$R7.1135952@news-east.usenetserver.com>
X-Abuse-Info: Please be sure to forward a copy of ALL headers
X-Abuse-Info: Otherwise we will be unable to process your complaint properly
X-Complaints-To: support@usenetserver.com
NNTP-Posting-Date: Mon, 03 Jul 2000 04:17:08 EDT
Date: Mon, 03 Jul 2000 08:17:08 GMT
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news.maxwell.syr.edu!newsfeed.skycache.com!Cidera!63.211.125.72!cyclone2.usenetserver.com!news-out.usenetserver.com!cyclone1.usenetserver.com!cyclone1.usenetserver.com!news-east.usenetserver.com.POSTED!not-for-mail
Xref: chonsp.franklin.ch alt.folklore.computers:59054

In article <39604596.3ECE7EC0@tnglwood.demon.co.uk>,
Bob Billing (AKA Uncle Bob) <unclebob@tnglwood.demon.co.uk> wrote:
>Adam Sampson wrote:
>
>> If you were building a new free OS, what features would you like from
>> other (non-Unix) OSs? I'd be very interested to hear the answers; this
>
>I'd like to have a DCL clone shell. I found shell programming far easier
>on the VAX than on unix.

Were there features in DCL you miss in sh, or was it more of a holistic
thing?  I haven't written anything in DCL in more than ten years, and I
didn't have Unix experience then, so I can't remember.
-- 
<kragen@pobox.com>       Kragen Sitaker     <http://www.pobox.com/~kragen/>
The Internet stock bubble didn't burst on 1999-11-08.  Hurrah!
<URL:http://www.pobox.com/~kragen/bubble.html>
The power didn't go out on 2000-01-01 either.  :)

######

From: Charles Eicher <ceicher@inav.net>
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: 3 Jul 2000 02:09:45 -0700
Organization: http://extra.newsguy.com
Lines: 31
Message-ID: <8jpl8p$gi0@edrn.newsguy.com>
References: <873dlsdrci.fsf@cartman.azz.net> <39604596.3ECE7EC0@tnglwood.demon.co.uk>
NNTP-Posting-Host: p-918.newsdawg.com
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!newsfeeds.belnet.be!news.belnet.be!news-spur1.maxwell.syr.edu!news.maxwell.syr.edu!newsfeed.stanford.edu!pln-w!spln!dex!extra.newsguy.com!newsp.newsguy.com!edrn
Xref: chonsp.franklin.ch alt.folklore.computers:58998

Adam Sampson wrote:

> If you were building a new free OS, what features would you like from
> other (non-Unix) OSs?

Gee, about the only thing I miss are the old JCL/RJE features where you could
issue commands for the operator to go around pulling tapes and disk packs off
the shelf and mount them, put special paper in the printer, etc. You just
punched the commands on the cards, and the operators ran around like little
drones, doing just what you told them to. Sometimes. Our operators were kinda
stupid, always loading the wrong tapes or spindling, mutilating, and folding our
punch cards. We used to have a special job header card that you weren't supposed
to punch, but instead, filled out with written instructions. I recall times when
my job was fine, but had failed several times due to I/O errors (Idiot Operator)
so I'd get a big black magic marker and write one word on each card, something
like:

DON'T
FORGET
TO
LOAD
TAPE01
YOU
SCREWED
UP
LAST
TIME

This would usually scare the operator into paying attention, they'd pick up a
deck with the header card showing just one huge word, "DON'T"...

######

Newsgroups: alt.folklore.computers
From: Terry Kennedy <terry@gate.tmk.com>
Subject: Re: Features to save from old OSs
X-Complaints-To: Email abuse@spc.edu if this posting is inappropriate
User-Agent: tin/1.4.2-20000205 ("Possession") (UNIX) (BSD/OS/4.1 (i386))
NNTP-Posting-Host: gate.tmk.com
Organization: St. Peter's College, US
Message-ID: <Fx4EGq.4CE@spcuna.spc.edu>
References: <873dlsdrci.fsf@cartman.azz.net> <877lb4f18g.fsf@litterbox.meowing.net>
X-Trace: spcuna.spc.edu 962625337 4543 terry [204.141.35.61]
Date: Mon, 3 Jul 2000 11:55:38 GMT
Lines: 23
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news.maxwell.syr.edu!newsfeed.nyu.edu!newsmaster.cc.columbia.edu!ord-feed.news.verio.net!news.verio.net!news.spc.edu!not-for-mail
Xref: chonsp.franklin.ch alt.folklore.computers:58972

Extra Strength Flufferin <supersede@india.com> writes:
[A long-ish list of features]

  It sounds like you're describing VMS 8-)

  What else do you need/want to preserve, feature-wise, that isn't in VMS?

  Offhand, I can think of:

  1) Faster process creation
  2) Ability to compile/run common Unix-based freeware without a lot of
     porting

  The second one drags in a bunch of stuff no matter where you say it -
shells (so you can run configure scripts), a bunch of utilities, Unix-
style pipes, and "Unix semantics" for I/O. VMS loses big-time on the
last - files are files and sockets are sockets, and you can't select()
on combinations thereof. This manages to make the simplest things a
complete nightmare - consider "talk", which wants to monitor the keyboard
and a network connection.

	Terry Kennedy             http://www.tmk.com
        terry@tmk.com             Jersey City, NJ USA

######

From: dg@pearl.tao.co.uk (David Given)
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: Mon, 3 Jul 2000 13:08:38 +0100
Organization: I do not speak for anyone but myself, and barely that.
Lines: 22
Message-ID: <6ovpj8.6a6.ln@127.0.0.1>
References: <873dlsdrci.fsf@cartman.azz.net>
NNTP-Posting-Host: 212.19.67.123
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
X-Trace: 962652064 IIX5YQT0T437BD413C uk25.supernews.com
X-Complaints-To: newsabuse@remarq.com
X-Newsreader: knews 1.0b.1
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!nyc-news-feed1.bbnplanet.com!news.gtei.net!colt.net!newsfeed.icl.net!newsfeed.skycache.com!Cidera!63.211.125.72!cyclone2.usenetserver.com!news-out.usenetserver.com!traffic.uncensored-news.com!sn-xit-02!supernews.com!sn-inject-01!127.0.0.1!nobody
Xref: chonsp.franklin.ch alt.folklore.computers:59113

In article <873dlsdrci.fsf@cartman.azz.net>,
	Adam Sampson <azz@gnu.org> writes:
[...]
> If you were building a new free OS, what features would you like from
> other (non-Unix) OSs?

Persistant/non-persistant datastore unification. Why have this artifical
division between RAM and disc space? All it does is make things
complicated.

Total platform independence (universal binaries).

Linda-based IPC.

Heterogenous multiprocessing. Remote access to resources should be
transparent.

-- 
+- David Given ---------------McQ-+ "No, I won't go and see your therapist! I
|  Work: dg@tao-group.com         | won't even go and see mine!" --- Ally
|  Play: dgiven@iname.com         | MacBeal                                    
+- http://wired.st-and.ac.uk/~dg -+ 

######

From: "Bob Billing (AKA Uncle Bob)" <unclebob@tnglwood.demon.co.uk>
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: Mon, 03 Jul 2000 15:24:31 +0100
Message-ID: <3960A21F.916C66B7@tnglwood.demon.co.uk>
References: <873dlsdrci.fsf@cartman.azz.net> <39604596.3ECE7EC0@tnglwood.demon.co.uk> <8_X75.24474$R7.1135952@news-east.usenetserver.com>
NNTP-Posting-Host: tnglwood.demon.co.uk
X-NNTP-Posting-Host: tnglwood.demon.co.uk:158.152.132.30
X-Trace: news.demon.co.uk 962637040 nnrp-06:3426 NO-IDENT tnglwood.demon.co.uk:158.152.132.30
X-Complaints-To: abuse@demon.net
X-Mailer: Mozilla 4.72 [en] (X11; U; Linux 2.2.14-5.0 i586)
X-Accept-Language: en
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Lines: 16
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news-fra1.dfn.de!news0.de.colt.net!colt.net!dispose.news.demon.net!demon!news.demon.co.uk!demon!tnglwood.demon.co.uk!falstaff.tanglewood!nobody
Xref: chonsp.franklin.ch alt.folklore.computers:59024

Kragen Sitaker wrote:

> Were there features in DCL you miss in sh, or was it more of a holistic
> thing?  I haven't written anything in DCL in more than ten years, and I
> didn't have Unix experience then, so I can't remember.

 It wasn't so much features as for some reason I found the syntax far
easier to memorise. Of course the grey cells were a lot younger then.
But it didn't seem to have as many odd special cases, and different ways
of doing the same thing, as for example bash.

-- 
I am Robert Billing, Christian, inventor, traveller, cook and animal
lover, I live near 0:46W 51:22N.  http://www.tnglwood.demon.co.uk/
"It burned me from within. It quickened; I was with book as a woman
is with child." CS Lewis - Till we have faces, Ch 21.

######

From: "P Linnane" <plyons@ca-unlimited.com>
Newsgroups: alt.folklore.computers
References: <873dlsdrci.fsf@cartman.azz.net> <877lb4f18g.fsf@litterbox.meowing.net> <Fx4EGq.4CE@spcuna.spc.edu>
Subject: Re: Features to save from old OSs
Lines: 14
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.00.2919.6600
X-MimeOLE: Produced By Microsoft MimeOLE V5.00.2919.6600
Organization: Ye 'Ol Disorganized NNTPCache groupie
Message-ID: <962652343.457300@server16.cable.com>
Cache-Post-Path: server16.cable.com!unknown@vp1-6.xcessible.com
X-Cache: nntpcache 2.4.0b2 (see http://www.nntpcache.org/)
Date: Mon, 3 Jul 2000 15:23:54 -0400
NNTP-Posting-Host: 204.92.64.17
X-Trace: nnrp1.uunet.ca 962652175 204.92.64.17 (Mon, 03 Jul 2000 15:22:55 EDT)
NNTP-Posting-Date: Mon, 03 Jul 2000 15:22:55 EDT
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news-fra1.dfn.de!news-koe1.dfn.de!RRZ.Uni-Koeln.DE!news.netcologne.de!newsfeed01.sul.t-online.de!newsfeed00.sul.t-online.de!t-online.de!diablo.theplanet.net!europa.netcrusader.net!206.191.82.230!prairie.attcanada.net!newsfeed.attcanada.net!142.77.1.188!news.uunet.ca!nnrp1.uunet.ca.POSTED!not-for-mail
Xref: chonsp.franklin.ch alt.folklore.computers:59080


"Terry Kennedy" <terry@gate.tmk.com> wrote in message
news:Fx4EGq.4CE@spcuna.spc.edu...
> Extra Strength Flufferin <supersede@india.com> writes:
> [A long-ish list of features]
>
>   It sounds like you're describing VMS 8-)
>

see figure 1.

######

From: Werner Uelpenich <werner.uelpenich@netcologne.de>
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: Mon, 03 Jul 2000 22:47:12 +0200
Lines: 32
Distribution: world
Message-ID: <3960FBD0.2C54E741@netcologne.de>
References: <873dlsdrci.fsf@cartman.azz.net> <6ovpj8.6a6.ln@127.0.0.1>
NNTP-Posting-Host: dial-194-8-209-171.netcologne.de
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Trace: news.netcologne.de 962657218 15049 194.8.209.171 (3 Jul 2000 20:46:58 GMT)
X-Complaints-To: abuse@netcologne.de
NNTP-Posting-Date: 3 Jul 2000 20:46:58 GMT
X-Mailer: Mozilla 4.7 [de]C-CCK-MCD QXW0322b  (WinNT; U)
X-Accept-Language: de,en
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news.maxwell.syr.edu!news.netcologne.de!not-for-mail
Xref: chonsp.franklin.ch alt.folklore.computers:59074



David Given schrieb:

> In article <873dlsdrci.fsf@cartman.azz.net>,
>         Adam Sampson <azz@gnu.org> writes:
> [...]
> > If you were building a new free OS, what features would you like from
> > other (non-Unix) OSs?
>
> Persistant/non-persistant datastore unification. Why have this artifical
> division between RAM and disc space? All it does is make things
> complicated.
>

Kent did it with K90 process control system, RAM space and disc space were called
"files" and organized in directories.

>
> Total platform independence (universal binaries).
>
> Linda-based IPC.
>
> Heterogenous multiprocessing. Remote access to resources should be
> transparent.
>
> --
> +- David Given ---------------McQ-+ "No, I won't go and see your therapist! I
> |  Work: dg@tao-group.com         | won't even go and see mine!" --- Ally
> |  Play: dgiven@iname.com         | MacBeal
> +- http://wired.st-and.ac.uk/~dg -+

######

From: Werner Uelpenich <werner.uelpenich@netcologne.de>
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: Mon, 03 Jul 2000 22:52:56 +0200
Lines: 27
Distribution: world
Message-ID: <3960FD28.889A6941@netcologne.de>
References: <873dlsdrci.fsf@cartman.azz.net> <39604596.3ECE7EC0@tnglwood.demon.co.uk> <8_X75.24474$R7.1135952@news-east.usenetserver.com> <3960A21F.916C66B7@tnglwood.demon.co.uk>
NNTP-Posting-Host: dial-194-8-209-171.netcologne.de
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Trace: news.netcologne.de 962657562 15468 194.8.209.171 (3 Jul 2000 20:52:42 GMT)
X-Complaints-To: abuse@netcologne.de
NNTP-Posting-Date: 3 Jul 2000 20:52:42 GMT
X-Mailer: Mozilla 4.7 [de]C-CCK-MCD QXW0322b  (WinNT; U)
X-Accept-Language: de,en
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!newscore.univie.ac.at!194.25.134.126.MISMATCH!newsfeed01.sul.t-online.de!t-online.de!news.netcologne.de!not-for-mail
Xref: chonsp.franklin.ch alt.folklore.computers:59072



"Bob Billing (AKA Uncle Bob)" schrieb:

> Kragen Sitaker wrote:
>
> > Were there features in DCL you miss in sh, or was it more of a holistic
> > thing?  I haven't written anything in DCL in more than ten years, and I
> > didn't have Unix experience then, so I can't remember.
>
>  It wasn't so much features as for some reason I found the syntax far
> easier to memorise. Of course the grey cells were a lot younger then.
> But it didn't seem to have as many odd special cases, and different ways
> of doing the same thing, as for example bash.
>

In unix shells you call a lot of small utilities with cryptic short names, in
DCL you call a function by a well memorizable name undependant from the
taskname which does the work at last.

>
> --
> I am Robert Billing, Christian, inventor, traveller, cook and animal
> lover, I live near 0:46W 51:22N.  http://www.tnglwood.demon.co.uk/
> "It burned me from within. It quickened; I was with book as a woman
> is with child." CS Lewis - Till we have faces, Ch 21.

######

From: wish@dumain.demon.co.uk (Bill Hay)
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: 3 Jul 2000 21:51:33 GMT
Organization: Dumain
Message-ID: <slrn8m22n5.akr.wish@dumain.demon.co.uk>
References: <873dlsdrci.fsf@cartman.azz.net>
NNTP-Posting-Host: sound.dumain.com
X-NNTP-Posting-Host: harkonnen.caladan.net:194.217.13.120
X-Trace: news.demon.co.uk 962678469 nnrp-09:21633 NO-IDENT harkonnen.caladan.net:194.217.13.120
X-Complaints-To: abuse@demon.net
NNTP-Posting-Date: 3 Jul 2000 21:51:33 GMT
User-Agent: slrn/0.9.5.6 (UNIX)
Lines: 10
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!serra.unipi.it!hammer.uoregon.edu!newsfeed.direct.ca!look.ca!newsfeeds.belnet.be!news.belnet.be!colt.net!dispose.news.demon.net!demon!news.demon.co.uk!demon!harkonnen.caladan.net!news.caladan.net!dumain.com!wish
Xref: chonsp.franklin.ch alt.folklore.computers:59110

In article <873dlsdrci.fsf@cartman.azz.net>, Adam Sampson wrote:
>If you were building a new free OS, what features would you like from
>other (non-Unix) OSs? I'd be very interested to hear the answers; this
Something I saw in CPM.  Shell builtins that were simple versions
of external commands and would invoke the external command if it
was passed an option it didn't understand. 


-- 
TANSTAAFM

######

From: Juergen Nickelsen <jnickelsen@acm.org>
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: 04 Jul 2000 00:45:37 +0200
Organization: [Posted via] Interactive Networx
Lines: 26
Sender: ni@goting.jn.berlin.snafu.de
Message-ID: <x7lmzizvce.fsf@goting.jn.berlin.snafu.de>
References: <873dlsdrci.fsf@cartman.azz.net> <39604596.3ECE7EC0@tnglwood.demon.co.uk> <8_X75.24474$R7.1135952@news-east.usenetserver.com>
NNTP-Posting-Host: ip184.berlin63.pub-ip.de.psi.net
X-Newsreader: Gnus v5.6.45/XEmacs 21.1 - "Canyonlands"
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!nyc-news-feed1.bbnplanet.com!news.gtei.net!colt.net!newspeer.clara.net!news.clara.net!unlisys!news.snafu.de!goting.jn.berlin.snafu.de!nobody
Xref: chonsp.franklin.ch alt.folklore.computers:59126

kragen@dnaco.net (Kragen Sitaker) writes:

> Were there features in DCL you miss in sh, or was it more of a
> holistic thing?

I have mixed feelings about DCL. The pipe-orientedness of the Unix
sh (and other utilities to make use of it) is a great thing, and an
area were DCL is awkward. (I have heard that this has changed in the
last 10 years, so please stop the beating.)

But what I really miss from DCL are the lexicals -- a *single* place
where you can get all information about the attributes of your
process and system and whatever. Unix (which I dearly love in spite
of all its weaknesses) offers you lots of commands to get at this
kind of information, all with different command syntax and output
format. The positive side of this is that it offers a challenge. :-}

(It would not be fair to even mention that some of these commands
are quite different in behaviour from vendor to vendor, or that some
might not exist on someone else's Unix, so that you have to employ a
completely different program to get at some piece of information if
you get it at all -- VMS is a proprietary OS with no such problems,
but another set of problems just because of that.)

-- 
Juergen Nickelsen

######

Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
References: <873dlsdrci.fsf@cartman.azz.net> <slrn8m22n5.akr.wish@dumain.demon.co.uk> <39615dcd@news.sierratel.com>
Organization: None
X-Newsreader: trn 4.0-test72 (19 April 1999)
From: kragen@dnaco.net (Kragen Sitaker)
Lines: 19
Message-ID: <v0d85.4324$R12.291365@news-east.usenetserver.com>
X-Abuse-Info: Please be sure to forward a copy of ALL headers
X-Abuse-Info: Otherwise we will be unable to process your complaint properly
X-Complaints-To: support@usenetserver.com
NNTP-Posting-Date: Mon, 03 Jul 2000 23:40:11 EDT
Date: Tue, 04 Jul 2000 03:40:11 GMT
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!isdnet!newsfeed.cwix.com!newsfeed.skycache.com!Cidera!63.211.125.72!cyclone2.usenetserver.com!news-out.usenetserver.com!cyclone1.usenetserver.com!cyclone1.usenetserver.com!news-east.usenetserver.com.POSTED!not-for-mail
Xref: chonsp.franklin.ch alt.folklore.computers:59128

In article <39615dcd@news.sierratel.com>,
Dowe Keller <dowe@krikkit.localdomain> wrote:
>On 3 Jul 2000 21:51:33 GMT, Bill Hay <wish@dumain.demon.co.uk> wrote:
>>In article <873dlsdrci.fsf@cartman.azz.net>, Adam Sampson wrote:
>>>If you were building a new free OS, what features would you like from
>>>other (non-Unix) OSs? I'd be very interested to hear the answers; this
>>Something I saw in CPM.  Shell builtins that were simple versions
>>of external commands and would invoke the external command if it
>>was passed an option it didn't understand. 
>
>IIRC some Unix shells do this same thing with the ls command.

tcsh, specifically, with ls-F, which invokes the command 'ls' to
process flags it doesn't recognize.
-- 
<kragen@pobox.com>       Kragen Sitaker     <http://www.pobox.com/~kragen/>
The Internet stock bubble didn't burst on 1999-11-08.  Hurrah!
<URL:http://www.pobox.com/~kragen/bubble.html>
The power didn't go out on 2000-01-01 either.  :)

######

From: korpela@ellie.ssl.berkeley.edu (Eric J. Korpela)
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: 4 Jul 2000 03:44:15 GMT
Organization: Cal Berkeley-- Space Sciences Lab
Lines: 28
Message-ID: <8jrmif$2l0$1@agate.berkeley.edu>
References: <873dlsdrci.fsf@cartman.azz.net>
NNTP-Posting-Host: ellie.ssl.berkeley.edu
X-Trace: agate.berkeley.edu 962682255 2720 128.32.18.173 (4 Jul 2000 03:44:15 GMT)
X-Complaints-To: abuse@berkeley.edu
NNTP-Posting-Date: 4 Jul 2000 03:44:15 GMT
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news.maxwell.syr.edu!news-hog.berkeley.edu!ucberkeley!agate.berkeley.edu!agate!ellie.ssl.berkeley.edu!korpela
Xref: chonsp.franklin.ch alt.folklore.computers:59091

In article <873dlsdrci.fsf@cartman.azz.net>, Adam Sampson  <azz@gnu.org> wrote:
>                             block-orientated IO for database
>operations

Sheesh, this is the new century and we're still doing explicit I/O to 
storage devices?  Either the memory you're accessing is backed by permanent
storage or its not.  The only difficulty is when you need to exceed your
address space.  Block oriented explicit I/O can be written on top of this
if necessary for legacy application support.

What databases need is not block-oriented I/O, but a means of serializing
memory access across multiple processes running on multiple systems.  The
VM system should have paging modes that chose serialized/non-serialized reads,
serialized/non-serialized writes and automatic update.  The system for which 
the storage is local needs to handle the serialization.  That means it needs 
to be in the network protocol that handles I/O to storage devices.  For 
example, if a system has mapped serialized a page for which the physical 
storage is managed by another system and the page is in local volatile storage, the local page will be invalidated by a memory write to the same page by
any other system.  For those familiar with unix mmap(), it means that
there should be another mapping mode MAP_SERIAL, which is the equivalent of
MAP_SHARED working network wide with automatic access control semaphores.
 
Eric
-- 
Eric Korpela                        |  An object at rest can never be
korpela@ssl.berkeley.edu            |  stopped.
<a href="http://sag-www.ssl.berkeley.edu/~korpela">Click for home page.</a>

######

From: dowe@krikkit.localdomain (Dowe Keller)
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
References: <873dlsdrci.fsf@cartman.azz.net> <slrn8m22n5.akr.wish@dumain.demon.co.uk>
X-Newsreader: slrn (0.9.4.3 UNIX)
NNTP-Posting-Host: 20923419618.sierratel.com
X-Original-NNTP-Posting-Host: 20923419618.sierratel.com
Message-ID: <39615dcd@news.sierratel.com>
Date: 3 Jul 2000 20:45:17 -0700
X-Trace: 3 Jul 2000 20:45:17 -0700, 20923419618.sierratel.com
Organization: news.sierratel.com
Lines: 16
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!newsfeeds.belnet.be!news.belnet.be!news-spur1.maxwell.syr.edu!news.maxwell.syr.edu!newsfeed.skycache.com!Cidera!209.155.26.10!news.sierratel.com!dowe
Xref: chonsp.franklin.ch alt.folklore.computers:59077

On 3 Jul 2000 21:51:33 GMT, Bill Hay <wish@dumain.demon.co.uk> wrote:
>In article <873dlsdrci.fsf@cartman.azz.net>, Adam Sampson wrote:
>>If you were building a new free OS, what features would you like from
>>other (non-Unix) OSs? I'd be very interested to hear the answers; this
>Something I saw in CPM.  Shell builtins that were simple versions
>of external commands and would invoke the external command if it
>was passed an option it didn't understand. 

IIRC some Unix shells do this same thing with the ls command.

-- 
dowe@sierratel.com
---
Real software engineers don't debug programs, they verify correctness.
This process doesn't necessarily involve execution of anything on a
computer, except perhaps a Correctness Verification Aid package.

######

From: "P Linnane" <plyons@ca-unlimited.com>
Newsgroups: alt.folklore.computers
References: <873dlsdrci.fsf@cartman.azz.net> <slrn8m22n5.akr.wish@dumain.demon.co.uk>
Subject: Re: Features to save from old OSs
Lines: 12
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.00.2919.6600
X-MimeOLE: Produced By Microsoft MimeOLE V5.00.2919.6600
Organization: Ye 'Ol Disorganized NNTPCache groupie
Message-ID: <962707831.948244@server16.cable.com>
Cache-Post-Path: server16.cable.com!unknown@vp1-3.xcessible.com
X-Cache: nntpcache 2.4.0b2 (see http://www.nntpcache.org/)
Date: Tue, 4 Jul 2000 06:48:39 -0400
NNTP-Posting-Host: 204.92.64.17
X-Trace: nnrp1.uunet.ca 962707668 204.92.64.17 (Tue, 04 Jul 2000 06:47:48 EDT)
NNTP-Posting-Date: Tue, 04 Jul 2000 06:47:48 EDT
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!newscore.univie.ac.at!194.25.134.126.MISMATCH!newsfeed01.sul.t-online.de!newsfeed00.sul.t-online.de!t-online.de!newsfeeds.belnet.be!news.belnet.be!news-spur1.maxwell.syr.edu!news.maxwell.syr.edu!sunqbc.risq.qc.ca!news.uunet.ca!nnrp1.uunet.ca.POSTED!not-for-mail
Xref: chonsp.franklin.ch alt.folklore.computers:59081


"Bill Hay" <wish@dumain.demon.co.uk> wrote in message > --

> TANSTAAFM

M?

"tanstaafl !" - RAH

######

From: multics@multics.ruserved.com
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: 4 Jul 2000 10:54:46 -0400
Organization: X-no-Archive: yes
Lines: 67
Message-ID: <8jstrm$8fd$1@multics.ruserved.com>
References: <873dlsdrci.fsf@cartman.azz.net>
NNTP-Posting-Host: ruserved.com
X-Original-NNTP-Posting-Host: ruserved.com
X-Trace: 4 Jul 2000 10:43:32 -0400, ruserved.com
XPident: multics
X-Original-NNTP-Posting-Host: 199.181.141.3
XPident: news
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!newsfeeds.belnet.be!news.belnet.be!newsfeed00.sul.t-online.de!t-online.de!diablo.theplanet.net!diablo.netcom.net.uk!netcom.net.uk!newsfeed.wizvax.net!news.wizvax.net!localhost!not-for-mail
Xref: chonsp.franklin.ch alt.folklore.computers:59100

In article <873dlsdrci.fsf@cartman.azz.net>, Adam Sampson  <azz@gnu.org> wrote:
[snip what did you like about other OS's]

I liked alot of things from Multics.  Its been too many years so I may
be explaining some things a little wrong, I'm sure the Multicians with
better memory and/or current experience will correct me.

1)  Access Control lists that were botheasy to manipulate and worked well.

2)  No 'file system', everything accessed as named virtual memory space.
When you referenced a 'file' (Multics term segment) on disk, it was mapped
into virtual memory automatically and referenced as virtual memory.
File I/O was simulated in virtual memory if you did file I/O to a
disk file.

3)  Almost no fatal errors in a program, an error pushed the entire
process on the stack and called in a new shell that could access the
the pushed process.  Using symbolic debugger you could examine and/or
change any variable and if it was a simple to fix problem, ie forgot to
initialize a variable, you could set the variable and resume execution
with the failed operation now fixed (ie divide by zero).

4)  The entire system is dynamic linked from the OS out by default,
none of the Unix kludges.  If you forgot to compile/write a subroutine,
you'd get a recoverable error (see e above).  Just write the routine,
compile it, and restart the failed call.

5)  Documantation that was very well maintained and very useful.
The Multics help files are the best online help files I've ever seen.
The unix man pages are often cryptic, incomplete, and sometimes contain
errors.

6)  Consistancy.  Everything worked the same 'way'.  If you had two
system routines with a 'long' vs 'brief' output option then the command
line options were always -l or -long for long.

7)  Multiple names for a segment (not links).  Multics allowed all
segments to have multiple names.  An example of this might be a math
library where the primary name was mathlib, but added names (additional
entry points) might be sin, cos, tan, etc.  Some commands might have a
long name and a short name.

8)  Instead of command line globbing, each command did its own globbing,
but there was globing support subroutines.  Its been many years to this
example may be wrong but your could copy and rename files at the same
time, ie copy *.c save/=.c.old would copy and change the extension on
all the files.

9)  Why does unix make compiling and running programs so complicated?
Multics had defualt search rules (user changable) for finding libraries
so a compile links and run would be:

	pl1 program  [system assume a .pl1 extension on the source]
	program      [run the compiled program, libraries dynamically linked as
	             [when referenced.]

note that the output from the compiler is the actual runnable code,
one of the benefits of memory being mapped to disk for persistant storage.

10)  every user got their own private process directory for all the temp
files used by the programs they run, no system /tmp shared by everyone.


For those looking for more information or to figure out where my errors
are, see http://www.multicians.org/ There are still Multics systems in
use today.  IIRC the first Multics system was booted in 1969.

######

Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
References: <873dlsdrci.fsf@cartman.azz.net>
Organization: Plethora . Net - More Net, Less Spam!
X-Newsreader: trn 4.0-test72 (19 April 1999)
From: seebs@plethora.net (Peter Seebach)
Date: 04 Jul 2000 15:52:32 GMT
Lines: 25
Message-ID: <39620840$0$8301$3c090ad1@news.plethora.net>
NNTP-Posting-Host: 85365622.news.plethora.net
X-Trace: 962725952 gemini.plethora.net 8301 seebs@205.166.146.8
X-Complaints-To: abuse@plethora.net
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!news.datacomm.ch!newsmaster-01.atnet.at!atnet.at!newscore.univie.ac.at!194.25.134.126.MISMATCH!newsfeed01.sul.t-online.de!t-online.de!unlisys!news.snafu.de!newsfeedZ.netscum.dQ!netscum.int!hermes.visi.com!news-out.visi.com!gemini-int.visi.com.MISMATCH!gemini.plethora.net.POSTED!not-for-mail
Xref: chonsp.franklin.ch alt.folklore.computers:59099

In article <873dlsdrci.fsf@cartman.azz.net>, Adam Sampson  <azz@gnu.org> wrote:
>If you were building a new free OS, what features would you like from
>other (non-Unix) OSs?

I want the Amiga's message-passing and IO systems.  Without the BPTR's.

Asynch I/O is a cool thing; it allows you to write a lot of programs more
expressively, and you can always provide synchronous I/O if you want it.

Their windowing system was actually pretty incredible, too; the OS did all
the menuing for you, if you wanted, and the *only* thing you needed to do
was:

1.  Hand it a list of menus.
2.  Wait for messages telling you what items had been selected.

It did multiple selections from a menu, promising to deliver them in order,
it redrew the screen automatically in the default case, and everything just
*worked*.

-s
-- 
Copyright 2000, All rights reserved.  Peter Seebach / seebs@plethora.net
C/Unix wizard, Pro-commerce radical, Spam fighter.  Boycott Spamazon!
Consulting & Computers: http://www.plethora.net/

######

From: jmfbahciv@aol.com
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: Thu, 06 Jul 00 10:34:30 GMT
Organization: UltraNet Communications, Inc.
Lines: 45
Message-ID: <8k21jt$r88$1@bob.news.rcn.net>
References: <873dlsdrci.fsf@cartman.azz.net> <025a9e75.2d4e185c@usw-ex0108-063.remarq.com> <8k1v41$mph$8@bob.news.rcn.net> <8k20d5$qd4$1@pegasus.csx.cam.ac.uk>
X-Trace: Ra01z4Z2jAOj4bT5/6pXJqieRvp9pb/HZKx5oMfQYC0=
X-Complaints-To: abuse@rcn.com
NNTP-Posting-Date: 6 Jul 2000 13:29:33 GMT
X-Newsreader:  News Xpress Version 1.0 Beta #4
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news.maxwell.syr.edu!feed1.news.rcn.net!rcn!207-172-255-172
Xref: chonsp.franklin.ch alt.folklore.computers:59224

In article <8k20d5$qd4$1@pegasus.csx.cam.ac.uk>,
   bjh21@cus.cam.ac.uk (Ben Harris) wrote:
>In article <8k1v41$mph$8@bob.news.rcn.net>,  <jmfbahciv@aol.com> wrote:
>>In article <025a9e75.2d4e185c@usw-ex0108-063.remarq.com>,
>>   martin2305 <m.taylorNOm.SPAM@elsevier.co.uk.invalid> wrote:
>>>Adam Sampson <azz@gnu.org> wrote:
>>>>If you were building a new free OS, what features would you
>>>like from
>>>>other (non-Unix) OSs?
>>>
>>>The ability for an executable image to possess file access
>>>privileges *in its own right* (I don't think UNIX OSs have this -
>>> please correct me if I'm mistaken). Many security problems
>>>occur because the user requires a generalised privilege to
>>>access data, when what is really required is the privilege to
>>>access the data *through an approved application*. A company's
>>>CFO, for example, while he may need update access to all
>>>financial data, needs that access to be mediated through an
>>>application, rather than being provided with the ability to
>>>patch the relevant data file in some ad-hoc way.
>>
>>JMF "upgraded" the TOPS-10 file protection mechanism to 
>>allow all kinds of stuff with his FILDAE (File Daemon).
>>Basically, if a certain file protection failure occurred,
>>instead of the monitor denying access, the whole smear was
>>handed to the file daemon to decide what access was allowed,
>>not allowed, and who could do either.
>
>That's cute.  Much like processors taking undefined instructions and page 
>fault traps -- it allows you to almost arbitrarily extend the semantics

With his design, it was essentially the owner of the file
who determined the type of access.  I don't know if his
spec is on-line; we had it in the Doc set so it might have
made it to the documentation tape that was finally shipped.
Also since the FILDAE was a user mode program that ran as
a system job, the customer site could write their own file
daemon to implement their own interpretations.  I don't think
many did; the one that Jim wrote seemed to satify the general
population.  I used the FILDAE extensively for source control
in our development shop.

/BAH

Subtract a hundred and four for e-mail.

######

From: jmfbahciv@aol.com
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: Fri, 07 Jul 00 08:55:09 GMT
Organization: UltraNet Communications, Inc.
Lines: 78
Message-ID: <8k4g5t$1ji$4@bob.news.rcn.net>
References: <873dlsdrci.fsf@cartman.azz.net> <962890329.979826@shelley.paradise.net.nz> <1ea9msoah4i5ehjg77bqjrv6i0corf4722@4ax.com> <FxArnK.401@spcuna.spc.edu> <962926887.864144@shelley.paradise.net.nz> <FxB09n.232@spcuna.spc.edu>
X-Trace: /cGMrzYWZZdac/dTVRiDjBrICcxYcFQtDjNNdUGgJus=
X-Complaints-To: abuse@rcn.com
NNTP-Posting-Date: 7 Jul 2000 11:50:21 GMT
X-Newsreader:  News Xpress Version 1.0 Beta #4
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!HSNX.atgi.net!news-hog.berkeley.edu!ucberkeley!news.maxwell.syr.edu!feed1.news.rcn.net!rcn!207-172-102-7
Xref: chonsp.franklin.ch alt.folklore.computers:59231

In article <FxB09n.232@spcuna.spc.edu>,
   Terry Kennedy <terry@gate.tmk.com> wrote:
>Don Stokes <don@news.daedalus.co.nz> writes:
>> Protected subsystems was the name I was thinking of.  It was going to 
>> be a new feature in V5, but didn't make it.  I thought I heard 
mutterings
>> about problems with protected subsystems in V6.somewthing -- hmmm I 
should
>> have a play with my 6.2 system and see if they work...
>
>  Yes, in early V6's there was an interaction between protected subsystems
>and sharable installed images. It was fixed by VAXSYS06_060 and in V6.1.
>
>> Still, if it made V6, then it did better than the grand plans for Phase
>> 5 DECnet (now known as DECnet/OSI).  Again, it was going to be out in
>> 5.0, but slipped and slipped -- it wasn't a standard component of 6.2,
>> although you ended up having to install it (usually accompanied by
>> extensive bad language) if you wanted some of the extra networking
>> products such as PSI (X.25) etc.  I think you can elect to stay with
>> Phase 4 in 7.x too.
>
>  Yup. There were some large DECnets which were pushing/exceeding the max-
>imum allowable amount of phase IV nodes, including DEC's own internal net.
>Rather than just increasing the address space, ...

I don't think that's [address space] the problem with the limit.
Lots of messes were created by the spec not allowing for 
extensibility and the notion that all Phases had to be treated
equally (rather than just supported).

> ...DEC embarked on a project
>to rewrite the whole product in a completely different 
>manner. While they were sinking in that tarpit, TCP/IP 
>(which had been mostly ignored* by VMS) ...

It may not have been "ignored" by the developers; it's just
that their development procedures were so clumsy that 
adjusting to the market was next to impossible.


>snuck up, and most of the external oversize DECnets solved their address
>space issues by moving to TCP/IP, either as a replacement or in addition
>to DECnet. 
>
>  When the product finally came out, people looked at it and barfed, most-
>ly due to the command interface and exceedingly verbose logging.

Even phase III and phase IV had that problem.

>
>  Since then, DEC has tried sneaking it in multiple ways:
>
>  1) Requiring it for things that previously worked 
>   under "classic" DECnet,
>     like PSI and the WAN Device Drivers.
>  2) Changing the product name (it's now DECnet-Plus).
>  3) Telling people scare stories about Phase IV (like requiring a special
>     item on your support contract at extra cost to get Phase IV support).
>
>  Despite all of this, there's a large "hell no, I won't go" 
>mindest in the customer base. All of the systems I deal with 
>are still using Phase IV, even
>on V7.2-1. If Phase IV is ever dropped, we'll deinstall DECnet rather than
>move to DECnet-Plus.

I'm confused.  I thought Palmer sold the DECnet business.

>
>  * Even the name of the DEC VMS TCP/IP package - "Ultrix Connection" or
>UCX for short - shows how much DEC missed the boat on the Internet - they
>thought people wanted TCP/IP to talk to their Ultrix boxes, not to use the
>Internet. Early UCX was missing NFS, the R-services, and so forth.

Or....it was only the Ultrix people who figured out the trends.
<snip>
/BAH

Subtract a hundred and four for e-mail.

######

From: jmfbahciv@aol.com
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: Thu, 06 Jul 00 09:51:54 GMT
Organization: UltraNet Communications, Inc.
Lines: 40
Message-ID: <8k1v41$mph$8@bob.news.rcn.net>
References: <873dlsdrci.fsf@cartman.azz.net> <025a9e75.2d4e185c@usw-ex0108-063.remarq.com>
X-Trace: u7gA4XpXmr2//v6TzKxhDSmRfT+0rrcQX9v5UPtxPxw=
X-Complaints-To: abuse@rcn.com
NNTP-Posting-Date: 6 Jul 2000 12:46:57 GMT
X-Newsreader:  News Xpress Version 1.0 Beta #4
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!isdnet!netnews.com!feed1.news.rcn.net!rcn!207-172-97-27
Xref: chonsp.franklin.ch alt.folklore.computers:59235

In article <025a9e75.2d4e185c@usw-ex0108-063.remarq.com>,
   martin2305 <m.taylorNOm.SPAM@elsevier.co.uk.invalid> wrote:
>Adam Sampson <azz@gnu.org> wrote:
>>If you were building a new free OS, what features would you
>like from
>>other (non-Unix) OSs?
>
>The ability for an executable image to possess file access
>privileges *in its own right* (I don't think UNIX OSs have this -
> please correct me if I'm mistaken). Many security problems
>occur because the user requires a generalised privilege to
>access data, when what is really required is the privilege to
>access the data *through an approved application*. A company's
>CFO, for example, while he may need update access to all
>financial data, needs that access to be mediated through an
>application, rather than being provided with the ability to
>patch the relevant data file in some ad-hoc way.

JMF "upgraded" the TOPS-10 file protection mechanism to 
allow all kinds of stuff with his FILDAE (File Daemon).
Basically, if a certain file protection failure occurred,
instead of the monitor denying access, the whole smear was
handed to the file daemon to decide what access was allowed,
not allowed, and who could do either.
 
>
>There, that feels better ;-).

Heh, heh.  Every once in a while you have do what makes you
feel better.

Personally, I think, at this point, I'd give a T-shirt for
buffered mode I/O.  No, I take that back.  Since I'm having
more and more software problems, I'd give just about
anything for the equivalent of a FILDDT.

/BAH


Subtract a hundred and four for e-mail.

######

From: jmfbahciv@aol.com
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: Sat, 08 Jul 00 08:07:17 GMT
Organization: UltraNet Communications, Inc.
Lines: 25
Message-ID: <8k71of$qdd$4@bob.news.rcn.net>
References: <873dlsdrci.fsf@cartman.azz.net> <025a9e75.2d4e185c@usw-ex0108-063.remarq.com> <87g0pmg9e8.fsf@cartman.azz.net>
X-Trace: YBH89uWkgEFpuU8r4XxWDJKoKdRjayIF4d9LKb+It28=
X-Complaints-To: abuse@rcn.com
NNTP-Posting-Date: 8 Jul 2000 11:02:39 GMT
X-Newsreader:  News Xpress Version 1.0 Beta #4
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news.maxwell.syr.edu!feed1.news.rcn.net!rcn!207-172-245-242
Xref: chonsp.franklin.ch alt.folklore.computers:59245

In article <87g0pmg9e8.fsf@cartman.azz.net>,
   Adam Sampson <azz@gnu.org> wrote:
>martin2305 <m.taylorNOm.SPAM@elsevier.co.uk.invalid> writes:
>
>> The ability for an executable image to possess file access
>> privileges *in its own right* (I don't think UNIX OSs have this -
>>  please correct me if I'm mistaken).
>
>Actually, Linux is currently implementing this, by way of adding a
>list of capabilites (Linux's implementation of privileges) to ELF
>executables. Presumably you need the setuid bit set to actually make
>it follow the suggested capabilities, and it means that you can't use
>the same trick with scripts.
>
>It seems to be agreed that this is only to be used as a kludge until
>the filesystem can store the capability set...

But you don't have to store it.  You can have the user define
access via a file.  The only change to the filesystem is
agreement on when the protection extension capability gets called.

/BAH


Subtract a hundred and four for e-mail.

######

From: jmfbahciv@aol.com
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: Fri, 07 Jul 00 09:03:10 GMT
Organization: UltraNet Communications, Inc.
Lines: 69
Message-ID: <8k4gku$1ji$5@bob.news.rcn.net>
References: <873dlsdrci.fsf@cartman.azz.net> <8jrmif$2l0$1@agate.berkeley.edu>
 <sit5msscksqbrr7o6kdetbd209t6267gsv@4ax.com> <8k1qvj$mph$1@bob.news.rcn.net> <rvc95.715$jE1.45272@typhoon.ne.mediaone.net>
X-Trace: rgepH71uv5f8ny9EQTKOqO4ThJXOxLV04FNmTQsiAqA=
X-Complaints-To: abuse@rcn.com
NNTP-Posting-Date: 7 Jul 2000 11:58:22 GMT
X-Newsreader:  News Xpress Version 1.0 Beta #4
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news.maxwell.syr.edu!feed1.news.rcn.net!rcn!207-172-102-7
Xref: chonsp.franklin.ch alt.folklore.computers:59249

In article <rvc95.715$jE1.45272@typhoon.ne.mediaone.net>,
   Ric Werme <werme@nospam.mediaone.net> wrote:
>jmfbahciv@aol.com writes:
>
>>In article <sit5msscksqbrr7o6kdetbd209t6267gsv@4ax.com>,
>>   Brian Inglis <Brian.dot.Inglis@SystematicSw.ab.ca> wrote:
>
>>>Some of use like to see our I/O move at high MB/s rates rather
>>>than low Mb/s rates. Like using Alpha servers with lots of
>>>engines and big memory -- to avoid the I/O as much as possible. 
>
>>[puzzled emoticon here] This sorta implies that I/O is being
>>used as a substitute for variables in the address space?
>>Isn't I/O supposed to be a repository of bits that one would
>>wish to carry over to a new machine?  I hadn't realized this
>>change of attitude before now.  
>
>Not really, the idea here is that if you have a big database, keep
>it on disk for safe-keeping, keep it in RAM for fast access.

That implies that there are data bases where all elements are
getting modified often.  I thought most databases consisted
of data that was mostly static.

>
>Celera Genomics has a DEC, err, Compaq Wildfire, err, GS160 system

That renaming of products is going to kill them faster than their
inability to make them.  You might pass that comment on.

>with 64 GB of RAM, which I think pretty much matches their library of
>human DNA fragments.  To assemble everything in the right order they
>have to compare a fragment with all the rest and look for matching
>stretches.  If they're lucky, a couple matches let them combine a few
>pieces together.

Obviously, this is an exception to the usual database rule.  
I really don't consider this a typical data base but more on
the order of a n-body problem. 
>
>It's a lot faster to do all this when the data is in RAM and stays there.
>So, fire up the assembly application, read umpteen GB of raw DNA
>sequences and start matching.

Right.  I guess people are using the term "data base" in a looser
way than when I was out there :-).  There was data base and
then there was computational.  

>
>See http://www.compaq.com/alphaserver/news/celera_june00.html
>
>Lessee, a RP02 disk was (is?) about 25 MB, IIRC.  So it would take
>over 2,500 (65,536/25) RP02 disk packs to hold all that data.
>It was only a few years ago that big Alpha servers crossed the 4 GB
>boundary, that we're up to 64 GB means that we're eating up addressing
>bits at a rate well in excess of Moore's Law.  Of course, no one will
>ever need more than 64 bits of addressing space, right?

Gee, is everybody inducing small computer thinking? ;-)
>
>The code is tiny, the OS is tiny.  Compile for speed, don't worry
>about code size.

Well, that's an attitude that I haven't understood.  (I used to
blame it on those COBOL programmers that didn't know any better.)

/BAH

Subtract a hundred and four for e-mail.

######

From: jmfbahciv@aol.com
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: Thu, 06 Jul 00 08:41:15 GMT
Organization: UltraNet Communications, Inc.
Lines: 94
Message-ID: <8k1qvj$mph$1@bob.news.rcn.net>
References: <873dlsdrci.fsf@cartman.azz.net> <8jrmif$2l0$1@agate.berkeley.edu> <sit5msscksqbrr7o6kdetbd209t6267gsv@4ax.com>
X-Trace: CQ5LBF4bGub9Xca+dX3CcBq6I0Q9V5wHdBtFxDqwhwE=
X-Complaints-To: abuse@rcn.com
NNTP-Posting-Date: 6 Jul 2000 11:36:19 GMT
X-Newsreader:  News Xpress Version 1.0 Beta #4
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!newsfeeds.belnet.be!news.belnet.be!news-spur1.maxwell.syr.edu!news.maxwell.syr.edu!feed1.news.rcn.net!rcn!207-172-97-27
Xref: chonsp.franklin.ch alt.folklore.computers:59267

In article <sit5msscksqbrr7o6kdetbd209t6267gsv@4ax.com>,
   Brian Inglis <Brian.dot.Inglis@SystematicSw.ab.ca> wrote:
>On 4 Jul 2000 03:44:15 GMT, korpela@ellie.ssl.berkeley.edu (Eric
>J. Korpela) wrote:
>
>>In article <873dlsdrci.fsf@cartman.azz.net>, Adam Sampson  <azz@gnu.org> 
wrote:
>>>                             block-orientated IO for database
>>>operations
>>
>>Sheesh, this is the new century and we're still doing explicit I/O to 
>>storage devices?  Either the memory you're accessing is backed by 
permanent
>>storage or its not.  The only difficulty is when you need to exceed your
>>address space.  Block oriented explicit I/O can be written on top of this
>>if necessary for legacy application support.
>>
>>What databases need is not block-oriented I/O, but a means of serializing
>>memory access across multiple processes running on multiple systems.  The
>
>Ugh -- shades of VMS clustering -- slowest I/O known to man --

From the little experience I had using VMS, all of its I/O was
slow.  That suggests that its approach to device access was not
efficient.

>databases need fast I/O to succeed -- serializing updates across
>multiple systems uses two phase commit -- with explicit
>recognition that it's faster if it's all on one system. 

It's not this simple.  The way databases get "fast" I/O is
via a sane way to solve contention caused by write.
>
>>VM system should have paging modes that chose serialized/non-serialized 
reads,
>>serialized/non-serialized writes and automatic update.  The system for 
which 
>>the storage is local needs to handle the serialization.  That means it 
needs 
>
>All memory should be local to the system, otherwise why pay the
>high price for fast access, and why bother with multiple systems
>(clusters) when you can just hang a bunch of engines, GBs of
>memory, and TBs of disk on one system, and use hierarchical MP
>locks within a single system. 

That didn't work very well and it was the main reason that
JMF and TW started the SMP project in TOPS-10.

>
>>to be in the network protocol that handles I/O to storage devices.  For 
>
>Obviously no need for performance if you're doing I/O over
>network protocols -- why not just use magtapes (and optional
>station wagons -- or minivans now?) 

You're getting silly now.  The reason for I/O over networks
is machines whose physical locations exceed the cable spec
limit.
>
>>example, if a system has mapped serialized a page for which the physical 
>>storage is managed by another system and the page is in local volatile 
storage,
>
>Why bother wasting memory if the storage is on another system
>across a network -- shared drives would be faster.

But you have a constraint of footprint allocation then. 
>
>>the local page will be invalidated by a memory write to the same page by
>>any other system.  For those familiar with unix mmap(), it means that
>>there should be another mapping mode MAP_SERIAL, which is the equivalent 
of
>>MAP_SHARED working network wide with automatic access control semaphores.
>
>Or just use cross system shared file locking if you have such
>meagre requirements. 

Barf.  I never liked that approach; but then I grew up under
TOPS-10 :-).

>Some of use like to see our I/O move at high MB/s rates rather
>than low Mb/s rates. Like using Alpha servers with lots of
>engines and big memory -- to avoid the I/O as much as possible. 

[puzzled emoticon here] This sorta implies that I/O is being
used as a substitute for variables in the address space?
Isn't I/O supposed to be a repository of bits that one would
wish to carry over to a new machine?  I hadn't realized this
change of attitude before now.  

/BAH

Subtract a hundred and four for e-mail.

######

From: Brian Inglis <Brian.dot.Inglis@SystematicSw.ab.ca>
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: Thu, 06 Jul 2000 01:12:05 -0600
Organization: Systematic Software
Reply-To: Brian.dot.Inglis@SystematicSw.ab.ca
Message-ID: <sit5msscksqbrr7o6kdetbd209t6267gsv@4ax.com>
References: <873dlsdrci.fsf@cartman.azz.net> <8jrmif$2l0$1@agate.berkeley.edu>
X-Newsreader: Forte Agent 1.8/32.548
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
NNTP-Posting-Host: 207.148.144.16
X-Original-NNTP-Posting-Host: 207.148.144.16
X-Trace: 6 Jul 2000 01:12:09 -0700, 207.148.144.16
Lines: 61
X-Original-NNTP-Posting-Host: 204.50.1.43
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news-fra1.dfn.de!news0.de.colt.net!colt.net!news.maxwell.syr.edu!nntp.flash.net!nntp.cadvision.com!news.cadvision.com!207.148.144.16
Xref: chonsp.franklin.ch alt.folklore.computers:59365

On 4 Jul 2000 03:44:15 GMT, korpela@ellie.ssl.berkeley.edu (Eric
J. Korpela) wrote:

>In article <873dlsdrci.fsf@cartman.azz.net>, Adam Sampson  <azz@gnu.org> wrote:
>>                             block-orientated IO for database
>>operations
>
>Sheesh, this is the new century and we're still doing explicit I/O to 
>storage devices?  Either the memory you're accessing is backed by permanent
>storage or its not.  The only difficulty is when you need to exceed your
>address space.  Block oriented explicit I/O can be written on top of this
>if necessary for legacy application support.
>
>What databases need is not block-oriented I/O, but a means of serializing
>memory access across multiple processes running on multiple systems.  The

Ugh -- shades of VMS clustering -- slowest I/O known to man --
databases need fast I/O to succeed -- serializing updates across
multiple systems uses two phase commit -- with explicit
recognition that it's faster if it's all on one system. 

>VM system should have paging modes that chose serialized/non-serialized reads,
>serialized/non-serialized writes and automatic update.  The system for which 
>the storage is local needs to handle the serialization.  That means it needs 

All memory should be local to the system, otherwise why pay the
high price for fast access, and why bother with multiple systems
(clusters) when you can just hang a bunch of engines, GBs of
memory, and TBs of disk on one system, and use hierarchical MP
locks within a single system. 

>to be in the network protocol that handles I/O to storage devices.  For 

Obviously no need for performance if you're doing I/O over
network protocols -- why not just use magtapes (and optional
station wagons -- or minivans now?) 

>example, if a system has mapped serialized a page for which the physical 
>storage is managed by another system and the page is in local volatile storage,

Why bother wasting memory if the storage is on another system
across a network -- shared drives would be faster. 

>the local page will be invalidated by a memory write to the same page by
>any other system.  For those familiar with unix mmap(), it means that
>there should be another mapping mode MAP_SERIAL, which is the equivalent of
>MAP_SHARED working network wide with automatic access control semaphores.

Or just use cross system shared file locking if you have such
meagre requirements. 
 
>Eric

Some of use like to see our I/O move at high MB/s rates rather
than low Mb/s rates. Like using Alpha servers with lots of
engines and big memory -- to avoid the I/O as much as possible. 

Thanks. Take care, Brian Inglis 	Calgary, Alberta, Canada
-- 
Brian_Inglis@CSi.com 	(Brian dot Inglis at SystematicSw dot ab dot ca)
				use address above to reply

######

X-Originating-Host: 193.131.222.100
Organization: http://www.remarq.com: The World's Usenet/Discussions Start Here
Subject: Re: Features to save from old OSs
Lines: 26
From: martin2305 <m.taylorNOm.SPAM@elsevier.co.uk.invalid>
Newsgroups: alt.folklore.computers
Message-ID: <025a9e75.2d4e185c@usw-ex0108-063.remarq.com>
References: <873dlsdrci.fsf@cartman.azz.net>
Bytes: 959
X-Wren-Trace: eGVAaGlwN302KD95YTxoY2VWcGxzZDN6ZXskZWM/aS4qMnw3JXQjOic2MzIldCYi
Date: Thu, 06 Jul 2000 04:11:26 -0700
NNTP-Posting-Host: 10.0.2.63
X-Complaints-To: wrenabuse@remarq.com
X-Trace: WReNphoon4 962882039 10.0.2.63 (Thu, 06 Jul 2000 04:13:59 PDT)
NNTP-Posting-Date: Thu, 06 Jul 2000 04:13:59 PDT
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news.maxwell.syr.edu!news.he.net!sn-xit-03!supernews.com!sn-inject-01!WReNclone!WReNphoon4.POSTED!WReN!not-for-mail
Xref: chonsp.franklin.ch alt.folklore.computers:59324

Adam Sampson <azz@gnu.org> wrote:
>If you were building a new free OS, what features would you
like from
>other (non-Unix) OSs?

The ability for an executable image to possess file access
privileges *in its own right* (I don't think UNIX OSs have this -
 please correct me if I'm mistaken). Many security problems
occur because the user requires a generalised privilege to
access data, when what is really required is the privilege to
access the data *through an approved application*. A company's
CFO, for example, while he may need update access to all
financial data, needs that access to be mediated through an
application, rather than being provided with the ability to
patch the relevant data file in some ad-hoc way.

There, that feels better ;-).

Martin Taylor


-----------------------------------------------------------

Got questions?  Get answers over the phone at Keen.com.
Up to 100 minutes free!
http://www.keen.com

######

From: "Bob Billing (AKA Uncle Bob)" <unclebob@tnglwood.demon.co.uk>
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: Thu, 06 Jul 2000 13:12:17 +0100
Message-ID: <396477A1.A772FA58@tnglwood.demon.co.uk>
References: <873dlsdrci.fsf@cartman.azz.net> <025a9e75.2d4e185c@usw-ex0108-063.remarq.com>
NNTP-Posting-Host: tnglwood.demon.co.uk
X-NNTP-Posting-Host: tnglwood.demon.co.uk:158.152.132.30
X-Trace: news.demon.co.uk 962886253 nnrp-10:28905 NO-IDENT tnglwood.demon.co.uk:158.152.132.30
X-Complaints-To: abuse@demon.net
X-Mailer: Mozilla 4.72 [en] (X11; U; Linux 2.2.14-5.0 i586)
X-Accept-Language: en
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Lines: 13
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!newscore.univie.ac.at!194.25.134.126.MISMATCH!newsfeed01.sul.t-online.de!newsfeed00.sul.t-online.de!t-online.de!diablo.theplanet.net!dispose.news.demon.net!demon!news.demon.co.uk!demon!tnglwood.demon.co.uk!falstaff.tanglewood!nobody
Xref: chonsp.franklin.ch alt.folklore.computers:59296

martin2305 wrote:

> The ability for an executable image to possess file access
> privileges *in its own right* (I don't think UNIX OSs have this -

 ISTR Maurice Wilkes telling me that the Titan operating system had
this.

-- 
I am Robert Billing, Christian, inventor, traveller, cook and animal
lover, I live near 0:46W 51:22N.  http://www.tnglwood.demon.co.uk/
"It burned me from within. It quickened; I was with book as a woman
is with child." CS Lewis - Till we have faces, Ch 21.

######

Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
References: <873dlsdrci.fsf@cartman.azz.net> <025a9e75.2d4e185c@usw-ex0108-063.remarq.com>
Organization: Daedalus Consulting
X-Newsreader: trn 4.0-test72 (19 April 1999)
From: don@news.daedalus.co.nz (Don Stokes)
Message-ID: <962890329.979826@shelley.paradise.net.nz>
Cache-Post-Path: shelley.paradise.net.nz!unknown@203-96-144-16.cable.paradise.net.nz
X-Cache: nntpcache 2.4.0b2 (see http://www.nntpcache.org/)
Lines: 14
Date: Thu, 06 Jul 2000 13:32:28 GMT
NNTP-Posting-Host: 203.96.152.26
X-Complaints-To: newsadmin@xtra.co.nz
X-Trace: news.xtra.co.nz 962890348 203.96.152.26 (Fri, 07 Jul 2000 01:32:28 NZST)
NNTP-Posting-Date: Fri, 07 Jul 2000 01:32:28 NZST
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!HSNX.atgi.net!feeder.via.net!newshub2.rdc1.sfba.home.com!news.home.com!enews.sgi.com!news.xtra.co.nz!not-for-mail
Xref: chonsp.franklin.ch alt.folklore.computers:59318

martin2305  <m.taylorNOm.SPAM@elsevier.co.uk.invalid> wrote:
>The ability for an executable image to possess file access
>privileges *in its own right* (I don't think UNIX OSs have this -
> please correct me if I'm mistaken). 

Unix sortof has this, via the setuid & setgid bits.  It's perhaps
not as general as you might like, but handles most cases moderately
well.

I seem to recall the ability to grant executable images access rights
identifiers as being a feature of VMS that was always going to be
available in the next release... did they ever get it to work?

-- don

######

From: korpela@sagan.ssl.berkeley.edu (Eric J. Korpela)
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: 6 Jul 2000 15:21:54 GMT
Organization: Cal Berkeley-- Space Sciences Lab
Lines: 90
Message-ID: <8k286i$jgh$1@agate.berkeley.edu>
References: <873dlsdrci.fsf@cartman.azz.net> <8jrmif$2l0$1@agate.berkeley.edu> <sit5msscksqbrr7o6kdetbd209t6267gsv@4ax.com>
NNTP-Posting-Host: sagan.ssl.berkeley.edu
X-Trace: agate.berkeley.edu 962896914 19985 128.32.18.166 (6 Jul 2000 15:21:54 GMT)
X-Complaints-To: abuse@berkeley.edu
NNTP-Posting-Date: 6 Jul 2000 15:21:54 GMT
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!News.Amsterdam.UnisourceCS!skynet.be!128.32.206.60.MISMATCH!news-hog.berkeley.edu!ucberkeley!agate.berkeley.edu!agate!sagan.ssl.berkeley.edu!korpela
Xref: chonsp.franklin.ch alt.folklore.computers:59278

In article <sit5msscksqbrr7o6kdetbd209t6267gsv@4ax.com>,
Brian Inglis  <Brian.dot.Inglis@SystematicSw.ab.ca> wrote:
>On 4 Jul 2000 03:44:15 GMT, korpela@ellie.ssl.berkeley.edu (Eric
>J. Korpela) wrote:
>
>>In article <873dlsdrci.fsf@cartman.azz.net>, Adam Sampson  <azz@gnu.org> wrote:
>>>                             block-orientated IO for database
>>>operations
>>
>>Sheesh, this is the new century and we're still doing explicit I/O to 
>>storage devices?  Either the memory you're accessing is backed by permanent
>>storage or its not.  The only difficulty is when you need to exceed your
>>address space.  Block oriented explicit I/O can be written on top of this
>>if necessary for legacy application support.
>>
>>What databases need is not block-oriented I/O, but a means of serializing
>>memory access across multiple processes running on multiple systems.  The
>
>Ugh -- shades of VMS clustering -- slowest I/O known to man --
>databases need fast I/O to succeed -- serializing updates across
>multiple systems uses two phase commit -- with explicit
>recognition that it's faster if it's all on one system. 

Which is why it's optional.  You don't do all I/O as serialized, you only
do specifically requested page swaps as serialized.  And there are times 
you can't do it all on one machine.  If you think otherwise you haven't 
done anything large.

>>serialized/non-serialized writes and automatic update.  The system for which 
>>the storage is local needs to handle the serialization.  That means it needs 
>
>All memory should be local to the system, otherwise why pay the
>high price for fast access, and why bother with multiple systems
>(clusters) when you can just hang a bunch of engines, GBs of
>memory, and TBs of disk on one system, and use hierarchical MP
>locks within a single system. 

1. It's cheaper.
2. Even in expensive systems there are only so many CPUs and
only so much memory you can add.  What happens when 8 or 16 or
256 CPUs are no longer enough?

>>to be in the network protocol that handles I/O to storage devices.  For 
>
>Obviously no need for performance if you're doing I/O over
>network protocols -- why not just use magtapes (and optional
>station wagons -- or minivans now?) 

Haven't noticed gigabit ethernet, have you?  Networks are getting
faster in case you haven't noticed.

>>example, if a system has mapped serialized a page for which the physical 
>>storage is managed by another system and the page is in local volatile storage,
>
>Why bother wasting memory if the storage is on another system
>across a network -- shared drives would be faster. 

What the hell does it matter what type of network it is?  Or don't
you think shared drives constitute a network?  Do you think there's
a need to arbitrate accesses to a shared drive?

>>the local page will be invalidated by a memory write to the same page by
>>any other system.  For those familiar with unix mmap(), it means that
>>there should be another mapping mode MAP_SERIAL, which is the equivalent of
>>MAP_SHARED working network wide with automatic access control semaphores.
>
>Or just use cross system shared file locking if you have such
>meagre requirements. 

In case you haven't noticed, it's a database.  You certainly don't want
to lock files, you want to lock the smallest addressable portion, i.e. the
page.  You do it in the VM system.  You write your explicit I/O to work
through the VM system.

>Some of use like to see our I/O move at high MB/s rates rather
>than low Mb/s rates. Like using Alpha servers with lots of
>engines and big memory -- to avoid the I/O as much as possible. 

Of course, that's why you memory map files.  But what happens when your
alpha server runs out of oomph?  There are only so many CPU slots.
How do prevent losing data in your database when the system crashes
if you aren't doing I/O?  How do you avoid I/O when your database is 
15 terabytes in size?  

Eric
-- 
Eric Korpela                        |  An object at rest can never be
korpela@ssl.berkeley.edu            |  stopped.
<a href="http://sag-www.ssl.berkeley.edu/~korpela">Click for home page.</a>

######

From: Alexandre Pechtchanski <alex*@*rockvax.rockefeller.edu>
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Organization: Rockefeller University Hospital (GCRC), New York
Message-ID: <1ea9msoah4i5ehjg77bqjrv6i0corf4722@4ax.com>
References: <873dlsdrci.fsf@cartman.azz.net> <025a9e75.2d4e185c@usw-ex0108-063.remarq.com> <962890329.979826@shelley.paradise.net.nz>
X-Newsreader: Forte Agent 1.7/32.534
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Lines: 21
Date: Thu, 06 Jul 2000 11:47:40 -0400
NNTP-Posting-Host: 129.85.24.56
X-Trace: rockyd.rockefeller.edu 962898625 129.85.24.56 (Thu, 06 Jul 2000 11:50:25 EDT)
NNTP-Posting-Date: Thu, 06 Jul 2000 11:50:25 EDT
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!News.Amsterdam.UnisourceCS!skynet.be!news-spur1.maxwell.syr.edu!news.maxwell.syr.edu!newsfeed.nyu.edu!rockyd.rockefeller.edu!not-for-mail
Xref: chonsp.franklin.ch alt.folklore.computers:59341

On Thu, 06 Jul 2000 13:32:28 GMT, don@news.daedalus.co.nz (Don Stokes) wrote:

>martin2305  <m.taylorNOm.SPAM@elsevier.co.uk.invalid> wrote:
>>The ability for an executable image to possess file access
>>privileges *in its own right* (I don't think UNIX OSs have this -
>> please correct me if I'm mistaken). 
>
>Unix sortof has this, via the setuid & setgid bits.  It's perhaps
>not as general as you might like, but handles most cases moderately
>well.
>
>I seem to recall the ability to grant executable images access rights
>identifiers as being a feature of VMS that was always going to be
>available in the next release... did they ever get it to work?

ISTR it was implemented in version 5.something.  At least you could install
privileged images and give unprivileged users run rights to them.

-- 
[ When replying, remove *'s from address ]
Alexandre Pechtchanski, Systems Manager, RUH, NY

######

From: strpic@linux.hr (Vid Strpic)
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: 6 Jul 2000 19:45:24 GMT
Organization: ELF is not Tolkien's
Lines: 16
Approved: The Croatian BOFH Team member and founder
Message-ID: <slrn8m9oek.f7k.strpic@lorien.elf.hr>
References: <873dlsdrci.fsf@cartman.azz.net> <slrn8m22n5.akr.wish@dumain.demon.co.uk> <962707831.948244@server16.cable.com>
NNTP-Posting-Host: dialin01-011.iskon.hr
Mime-Version: 1.0
Content-Type: text/plain; charset=iso-8859-2
Content-Transfer-Encoding: 8bit
X-Trace: sunce.iskon.hr 962914781 14851 213.191.129.20 (6 Jul 2000 20:19:41 GMT)
X-Complaints-To: abuse@iskon.hr
NNTP-Posting-Date: 6 Jul 2000 20:19:41 GMT
X-No-Archive: no
X-Attribution: Vid
X-Troll: no
User-Agent: slrn/0.9.6.2 (Linux)
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news-fra1.dfn.de!fu-berlin.de!news.maxwell.syr.edu!newsfeed.cwix.com!argos.tel.hr!Iskon!lorien!strpic
Xref: chonsp.franklin.ch alt.folklore.computers:59326

P Linnane said unto us in alt.folklore.computers :
>"Bill Hay" <wish@dumain.demon.co.uk> wrote in message > --
>> TANSTAAFM
>M?

Meal?

>"tanstaafl !" - RAH

Maybe he was just a bit generalistic.

-- 
 (o_      Vid Strpiæ, also known as Martin.  <strpic@linux.hr>      )) 
 //\        (I don't speak for my employer, just for myself.)      ((  
 V_/_                                                            C|~~| 
        UNIX fundamentalist - and an average chauvinistic male.   `--' 

######

Newsgroups: alt.folklore.computers
From: Terry Kennedy <terry@gate.tmk.com>
Subject: Re: Features to save from old OSs
X-Complaints-To: Email abuse@spc.edu if this posting is inappropriate
User-Agent: tin/1.4.2-20000205 ("Possession") (UNIX) (BSD/OS/4.1 (i386))
NNTP-Posting-Host: gate.tmk.com
Organization: St. Peter's College, US
Message-ID: <FxArnK.401@spcuna.spc.edu>
References: <873dlsdrci.fsf@cartman.azz.net> <025a9e75.2d4e185c@usw-ex0108-063.remarq.com> <962890329.979826@shelley.paradise.net.nz> <1ea9msoah4i5ehjg77bqjrv6i0corf4722@4ax.com>
X-Trace: spcuna.spc.edu 962922367 4935 terry [204.141.35.61]
Date: Thu, 6 Jul 2000 22:26:08 GMT
Lines: 17
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!isdnet!netnews.com!cpk-news-hub1.bbnplanet.com!crtntx1-snh1.gtei.net!news.gtei.net!dfw-peer.news.verio.net!ord-feed.news.verio.net!news.verio.net!news.spc.edu!not-for-mail
Xref: chonsp.franklin.ch alt.folklore.computers:59198

Alexandre Pechtchanski <alex*@*rockvax.rockefeller.edu> writes:
> On Thu, 06 Jul 2000 13:32:28 GMT, don@news.daedalus.co.nz (Don Stokes) wrote:
>>I seem to recall the ability to grant executable images access rights
>>identifiers as being a feature of VMS that was always going to be
>>available in the next release... did they ever get it to work?
>
> ISTR it was implemented in version 5.something.  At least you could install
> privileged images and give unprivileged users run rights to them.

  Images installed with privs have been around for ages (since at least 4.4,
when I started with VMS). I think Don was talking about "protected subsystems"
which finally shipped in VMS V6.0. For more info, see:
http://www.openvms.compaq.com/wizard/wiz_1547.html
http://www.openvms.compaq.com:8000/72final/6346/6346pro_contents_001.html#toc_chapter_13

	Terry Kennedy             http://www.tmk.com
        terry@tmk.com             Jersey City, NJ USA

######

Sender: azz@cartman.azz.net
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
References: <873dlsdrci.fsf@cartman.azz.net> <025a9e75.2d4e185c@usw-ex0108-063.remarq.com>
From: Adam Sampson <azz@gnu.org>
Message-ID: <87g0pmg9e8.fsf@cartman.azz.net>
Lines: 19
X-Newsreader: Gnus v5.6.45/XEmacs 21.1 - "20 Minutes to Nikko"
Date: 06 Jul 2000 23:51:27 +0100
NNTP-Posting-Host: 212.159.29.126
X-Complaints-To: abuse@plus.net.uk
X-Trace: stones 963001023 212.159.29.126 (Fri, 07 Jul 2000 21:17:03 BST)
NNTP-Posting-Date: Fri, 07 Jul 2000 21:17:03 BST
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!enews.sgi.com!newshub2.rdc1.sfba.home.com!news.home.com!newsfeed.direct.ca!look.ca!btnet-peer0!btnet!landlord!stones.POSTED!cartman.azz.net!nobody
Xref: chonsp.franklin.ch alt.folklore.computers:59283

martin2305 <m.taylorNOm.SPAM@elsevier.co.uk.invalid> writes:

> The ability for an executable image to possess file access
> privileges *in its own right* (I don't think UNIX OSs have this -
>  please correct me if I'm mistaken).

Actually, Linux is currently implementing this, by way of adding a
list of capabilites (Linux's implementation of privileges) to ELF
executables. Presumably you need the setuid bit set to actually make
it follow the suggested capabilities, and it means that you can't use
the same trick with scripts.

It seems to be agreed that this is only to be used as a kludge until
the filesystem can store the capability set...

-- 

Adam Sampson
azz@gnu.org

######

Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
References: <873dlsdrci.fsf@cartman.azz.net> <962890329.979826@shelley.paradise.net.nz> <1ea9msoah4i5ehjg77bqjrv6i0corf4722@4ax.com> <FxArnK.401@spcuna.spc.edu>
Organization: Daedalus Consulting
X-Newsreader: trn 4.0-test72 (19 April 1999)
From: don@news.daedalus.co.nz (Don Stokes)
Message-ID: <962926887.864144@shelley.paradise.net.nz>
Cache-Post-Path: shelley.paradise.net.nz!unknown@203-96-144-16.cable.paradise.net.nz
X-Cache: nntpcache 2.4.0b2 (see http://www.nntpcache.org/)
Lines: 31
Date: Thu, 06 Jul 2000 23:41:46 GMT
NNTP-Posting-Host: 203.96.152.26
X-Complaints-To: newsadmin@xtra.co.nz
X-Trace: news.xtra.co.nz 962926906 203.96.152.26 (Fri, 07 Jul 2000 11:41:46 NZST)
NNTP-Posting-Date: Fri, 07 Jul 2000 11:41:46 NZST
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!News.Amsterdam.UnisourceCS!skynet.be!newsfeed.stanford.edu!headwall.stanford.edu!feeder.via.net!enews.sgi.com!news.xtra.co.nz!not-for-mail
Xref: chonsp.franklin.ch alt.folklore.computers:59317

In article <FxArnK.401@spcuna.spc.edu>,
Terry Kennedy  <terry@gate.tmk.com> wrote:
>Alexandre Pechtchanski <alex*@*rockvax.rockefeller.edu> writes:
>> On Thu, 06 Jul 2000 13:32:28 GMT, don@news.daedalus.co.nz (Don Stokes) wrote:
>>>I seem to recall the ability to grant executable images access rights
>>>identifiers as being a feature of VMS that was always going to be
>>>available in the next release... did they ever get it to work?
>
>  Images installed with privs have been around for ages (since at least 4.4,
>when I started with VMS). I think Don was talking about "protected subsystems"
>which finally shipped in VMS V6.0. For more info, see:

I'm sure installed-with-privs was there from V1. 

Protected subsystems was the name I was thinking of.  It was going to 
be a new feature in V5, but didn't make it.  I thought I heard mutterings
about problems with protected subsystems in V6.somewthing -- hmmm I should
have a play with my 6.2 system and see if they work...

Still, if it made V6, then it did better than the grand plans for Phase
5 DECnet (now known as DECnet/OSI).  Again, it was going to be out in
5.0, but slipped and slipped -- it wasn't a standard component of 6.2,
although you ended up having to install it (usually accompanied by
extensive bad language) if you wanted some of the extra networking
products such as PSI (X.25) etc.  I think you can elect to stay with
Phase 4 in 7.x too.

In the more than ten years since Phase 5 was announced, OSI has simply
been and gone.

-- don

######

Newsgroups: alt.folklore.computers
From: Terry Kennedy <terry@gate.tmk.com>
Subject: Re: Features to save from old OSs
X-Complaints-To: Email abuse@spc.edu if this posting is inappropriate
User-Agent: tin/1.4.2-20000205 ("Possession") (UNIX) (BSD/OS/4.1 (i386))
NNTP-Posting-Host: gate.tmk.com
Organization: St. Peter's College, US
Message-ID: <FxB09n.232@spcuna.spc.edu>
References: <873dlsdrci.fsf@cartman.azz.net> <962890329.979826@shelley.paradise.net.nz> <1ea9msoah4i5ehjg77bqjrv6i0corf4722@4ax.com> <FxArnK.401@spcuna.spc.edu> <962926887.864144@shelley.paradise.net.nz>
X-Trace: spcuna.spc.edu 962933530 1144 terry [204.141.35.61]
Date: Fri, 7 Jul 2000 01:32:11 GMT
Lines: 56
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news.maxwell.syr.edu!nntp.flash.net!dfw-peer.news.verio.net!ord-feed.news.verio.net!news.verio.net!news.spc.edu!not-for-mail
Xref: chonsp.franklin.ch alt.folklore.computers:59190

Don Stokes <don@news.daedalus.co.nz> writes:
> Protected subsystems was the name I was thinking of.  It was going to 
> be a new feature in V5, but didn't make it.  I thought I heard mutterings
> about problems with protected subsystems in V6.somewthing -- hmmm I should
> have a play with my 6.2 system and see if they work...

  Yes, in early V6's there was an interaction between protected subsystems
and sharable installed images. It was fixed by VAXSYS06_060 and in V6.1.

> Still, if it made V6, then it did better than the grand plans for Phase
> 5 DECnet (now known as DECnet/OSI).  Again, it was going to be out in
> 5.0, but slipped and slipped -- it wasn't a standard component of 6.2,
> although you ended up having to install it (usually accompanied by
> extensive bad language) if you wanted some of the extra networking
> products such as PSI (X.25) etc.  I think you can elect to stay with
> Phase 4 in 7.x too.

  Yup. There were some large DECnets which were pushing/exceeding the max-
imum allowable amount of phase IV nodes, including DEC's own internal net.
Rather than just increasing the address space, DEC embarked on a project
to rewrite the whole product in a completely different manner. While they
were sinking in that tarpit, TCP/IP (which had been mostly ignored* by VMS)
snuck up, and most of the external oversize DECnets solved their address
space issues by moving to TCP/IP, either as a replacement or in addition
to DECnet. 

  When the product finally came out, people looked at it and barfed, most-
ly due to the command interface and exceedingly verbose logging.

  Since then, DEC has tried sneaking it in multiple ways:

  1) Requiring it for things that previously worked under "classic" DECnet,
     like PSI and the WAN Device Drivers.
  2) Changing the product name (it's now DECnet-Plus).
  3) Telling people scare stories about Phase IV (like requiring a special
     item on your support contract at extra cost to get Phase IV support).

  Despite all of this, there's a large "hell no, I won't go" mindest in the
customer base. All of the systems I deal with are still using Phase IV, even
on V7.2-1. If Phase IV is ever dropped, we'll deinstall DECnet rather than
move to DECnet-Plus.

  * Even the name of the DEC VMS TCP/IP package - "Ultrix Connection" or
UCX for short - shows how much DEC missed the boat on the Internet - they
thought people wanted TCP/IP to talk to their Ultrix boxes, not to use the
Internet. Early UCX was missing NFS, the R-services, and so forth.

> In the more than ten years since Phase 5 was announced, OSI has simply
> been and gone.

  Did it ever actually get to the point where someone could say it had
"been"? I recall it more as "it's coming! it's coming! it's coming! <blink>
what? never heard of it!"

	Terry Kennedy             http://www.tmk.com
        terry@tmk.com             Jersey City, NJ USA

######

Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
References: <873dlsdrci.fsf@cartman.azz.net> <8jrmif$2l0$1@agate.berkeley.edu>
     <sit5msscksqbrr7o6kdetbd209t6267gsv@4ax.com> <8k1qvj$mph$1@bob.news.rcn.net>
From: Ric Werme <werme@nospam.mediaone.net>
X-Newsreader: NN version 6.5.0 CURRENT #119
Lines: 45
Message-ID: <rvc95.715$jE1.45272@typhoon.ne.mediaone.net>
Date: Fri, 07 Jul 2000 03:53:59 GMT
NNTP-Posting-Host: 24.91.12.15
X-Complaints-To: abuse@mediaone.net
X-Trace: typhoon.ne.mediaone.net 962942039 24.91.12.15 (Thu, 06 Jul 2000 23:53:59 EDT)
NNTP-Posting-Date: Thu, 06 Jul 2000 23:53:59 EDT
Organization: Road Runner
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!isdnet!newsfeed.cwix.com!chnws02.mediaone.net!chnws05.ne.mediaone.net!24.128.8.70!typhoon.ne.mediaone.net.POSTED!not-for-mail
Xref: chonsp.franklin.ch alt.folklore.computers:59336

jmfbahciv@aol.com writes:

>In article <sit5msscksqbrr7o6kdetbd209t6267gsv@4ax.com>,
>   Brian Inglis <Brian.dot.Inglis@SystematicSw.ab.ca> wrote:

>>Some of use like to see our I/O move at high MB/s rates rather
>>than low Mb/s rates. Like using Alpha servers with lots of
>>engines and big memory -- to avoid the I/O as much as possible. 

>[puzzled emoticon here] This sorta implies that I/O is being
>used as a substitute for variables in the address space?
>Isn't I/O supposed to be a repository of bits that one would
>wish to carry over to a new machine?  I hadn't realized this
>change of attitude before now.  

Not really, the idea here is that if you have a big database, keep
it on disk for safe-keeping, keep it in RAM for fast access.

Celera Genomics has a DEC, err, Compaq Wildfire, err, GS160 system
with 64 GB of RAM, which I think pretty much matches their library of
human DNA fragments.  To assemble everything in the right order they
have to compare a fragment with all the rest and look for matching
stretches.  If they're lucky, a couple matches let them combine a few
pieces together.

It's a lot faster to do all this when the data is in RAM and stays there.
So, fire up the assembly application, read umpteen GB of raw DNA
sequences and start matching.

See http://www.compaq.com/alphaserver/news/celera_june00.html

Lessee, a RP02 disk was (is?) about 25 MB, IIRC.  So it would take
over 2,500 (65,536/25) RP02 disk packs to hold all that data.
It was only a few years ago that big Alpha servers crossed the 4 GB
boundary, that we're up to 64 GB means that we're eating up addressing
bits at a rate well in excess of Moore's Law.  Of course, no one will
ever need more than 64 bits of addressing space, right?

The code is tiny, the OS is tiny.  Compile for speed, don't worry
about code size.

	-Ric
--
Ric Werme                            | werme@nospam.mediaone.net
http://people.ne.mediaone.net/werme  |       ^^^^^^^ delete

######

Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
References: <873dlsdrci.fsf@cartman.azz.net> <sit5msscksqbrr7o6kdetbd209t6267gsv@4ax.com> <8k1qvj$mph$1@bob.news.rcn.net> <rvc95.715$jE1.45272@typhoon.ne.mediaone.net>
Organization: None
X-Newsreader: trn 4.0-test72 (19 April 1999)
From: kragen@dnaco.net (Kragen Sitaker)
Lines: 18
Message-ID: <d%c95.14639$N7.662850@news-east.usenetserver.com>
X-Abuse-Info: Please be sure to forward a copy of ALL headers
X-Abuse-Info: Otherwise we will be unable to process your complaint properly
X-Complaints-To: support@usenetserver.com
NNTP-Posting-Date: Fri, 07 Jul 2000 00:27:53 EDT
Date: Fri, 07 Jul 2000 04:27:53 GMT
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news.maxwell.syr.edu!newsfeed.skycache.com!Cidera!63.211.125.72!cyclone2.usenetserver.com!news-out.usenetserver.com!cyclone1.usenetserver.com!cyclone1.usenetserver.com!news-east.usenetserver.com.POSTED!not-for-mail
Xref: chonsp.franklin.ch alt.folklore.computers:59371

In article <rvc95.715$jE1.45272@typhoon.ne.mediaone.net>,
Ric Werme  <werme@nospam.mediaone.net> wrote:
>Lessee, a RP02 disk was (is?) about 25 MB, IIRC.  So it would take
>over 2,500 (65,536/25) RP02 disk packs to hold all that data.
>It was only a few years ago that big Alpha servers crossed the 4 GB
>boundary, that we're up to 64 GB means that we're eating up addressing
>bits at a rate well in excess of Moore's Law.  Of course, no one will
>ever need more than 64 bits of addressing space, right?
>
>The code is tiny, the OS is tiny.  Compile for speed, don't worry
>about code size.

The cache is tiny, too.  ;)
-- 
<kragen@pobox.com>       Kragen Sitaker     <http://www.pobox.com/~kragen/>
The Internet stock bubble didn't burst on 1999-11-08.  Hurrah!
<URL:http://www.pobox.com/~kragen/bubble.html>
The power didn't go out on 2000-01-01 either.  :)

######

From: wish@dumain.demon.co.uk (Bill Hay)
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: 7 Jul 2000 07:00:57 GMT
Organization: Dumain
Lines: 9
Message-ID: <slrn8mb019.6r7.wish@dumain.demon.co.uk>
References: <873dlsdrci.fsf@cartman.azz.net> <slrn8m22n5.akr.wish@dumain.demon.co.uk> <962707831.948244@server16.cable.com>
NNTP-Posting-Host: sound.dumain.com
X-Trace: dumain.demon.co.uk 962953257 2931 127.0.0.2 (7 Jul 2000 07:00:57 GMT)
X-Complaints-To: usenet@dumain.com
NNTP-Posting-Date: 7 Jul 2000 07:00:57 GMT
User-Agent: slrn/0.9.5.6 (UNIX)
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news.maxwell.syr.edu!newsfeed.stanford.edu!news.kjsl.com!drmach.dream.org.uk!drmach.demon.co.uk!dumain.com!wish
Xref: chonsp.franklin.ch alt.folklore.computers:59331

In article <962707831.948244@server16.cable.com>, P Linnane wrote:
>"Bill Hay" <wish@dumain.demon.co.uk> wrote in message > --
>> TANSTAAFM
>M?
Market


-- 
TANSTAAFM

######

From: bediger@csn.net (Bruce Ediger)
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Organization: Queen's Own East Essex Bicycle Fusiliers
Lines: 30
Message-ID: <8k583b$mgh@teal.sni.net>
References: <873dlsdrci.fsf@cartman.azz.net> <877lb4f18g.fsf@litterbox.meowing.net> <cNT75.4138$7%3.326535@news.flash.net>
Date: 7 Jul 2000 12:38:35 -0600
NNTP-Posting-Host: 199.117.27.22
X-Complaints-To: news-admin@qwestip.net
X-Trace: wdc-read-01.qwest.net 962995530 199.117.27.22 (Fri, 07 Jul 2000 12:45:30 MDT)
NNTP-Posting-Date: Fri, 07 Jul 2000 12:45:30 MDT
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!News.Amsterdam.UnisourceCS!skynet.be!cyclone.bc.net!HSNX.atgi.net!den-news-01.qwest.net!qwest!wdc-read-01.qwest.net!teal.csn.net!not-for-mail
Xref: chonsp.franklin.ch alt.folklore.computers:59185

"George R. Gonzalez" <grg1@flash.net> wrote:
>Some of the nice features from CDC's KRONOS and NOS operating systems.

	[...]

>*  Local files.   Each task's files were in their own namespace, and went
>away
>when the job ended, unless you explicitly attached a global file, or saved
>the local
>file to permanent storage (EXTREMELY double-edged).  But it meant much
>less muss and fuss cleaning up after a job was run.

This was perhaps the worst feature of NOS, particularly since each file
had a bizarre "type" associated with it, and the method of fetching
files into the local namespace was different for each of the types.
There may have only been two types, since I can only recall "ATTACH"
and "GET" as fetching methods.  Besides that, ATTACHed files went to
permanent storage automatically.  Even when you went through the ritual
of fetching them into the local namespace, the changes were permanent
even if you didn't save them.  This misfeature cost untold pain and
lost and overwritten data.

>There were quite a few misdesigns too, but let's not dwell on those...

The misdesigns in NOS included single-level directory structure,
limited length file names (10 character?), non-ASCII character set, 
and bizarrely over-complicated I/O scheme(s).
-- 
Once, galactic empires might have seemed a Post-Human domain.
Now, sadly, even interplanetary ones are.

######

From: rhawkins@iastate.edu (Rick Hawkins)
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: 7 Jul 2000 20:14:07 GMT
Organization: Unknown
Lines: 38
Message-ID: <8k5dmf$8lv$1@news.iastate.edu>
References: <873dlsdrci.fsf@cartman.azz.net> <39620840$0$8301$3c090ad1@news.plethora.net>
NNTP-Posting-Host: pv2079.vincent.iastate.edu
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!enews.sgi.com!logbridge.uoregon.edu!newsrelay.iastate.edu!news.iastate.edu!rhawkins
Xref: chonsp.franklin.ch alt.folklore.computers:59265

In article <39620840$0$8301$3c090ad1@news.plethora.net>,
Peter Seebach <seebs@plethora.net> wrote:
>In article <873dlsdrci.fsf@cartman.azz.net>, Adam Sampson  <azz@gnu.org> wrote:
>>If you were building a new free OS, what features would you like from
>>other (non-Unix) OSs?

>I want the Amiga's message-passing and IO systems.  Without the BPTR's.
...
>Their windowing system was actually pretty incredible, too; the OS did all
>the menuing for you, if you wanted, and the *only* thing you needed to do
>was:
>
>1.  Hand it a list of menus.
>2.  Wait for messages telling you what items had been selected.

I remember doing this on the mac, but I cannot for the life of me
remember what I was using.  Hypercard? Supercard? Foxbase? SOmething
else?

>It did multiple selections from a menu, promising to deliver them in order,
>it redrew the screen automatically in the default case, and everything just
>*worked*.

in order?  that's neat . . . 

hawk, thoroughly baffled by what he was doing

p.s. This is a posting test.  I pasted the whole bloody thing from
the machine where' I'm reading into the xwindow for the isu machine . . . 
I suppose I could write a script to watch for ftp'd files and post,
but that seems a bit extreme . . .


-- 
Richard E. Hawkins
dochawk@psu.edu [regardless of where the message says it comes from]

These opinions will not be those of  Penn State until they pay my retainer.

######

From: rhawkins@iastate.edu (Rick Hawkins)
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: 7 Jul 2000 20:53:07 GMT
Organization: Unknown
Lines: 19
Message-ID: <8k5fvj$8t8$1@news.iastate.edu>
References: <873dlsdrci.fsf@cartman.azz.net> <8_X75.24474$R7.1135952@news-east.usenetserver.com> <3960A21F.916C66B7@tnglwood.demon.co.uk> <3960FD28.889A6941@netcologne.de>
NNTP-Posting-Host: pv2079.vincent.iastate.edu
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!enews.sgi.com!feeder.via.net!headwall.stanford.edu!unlnews.unl.edu!newsrelay.iastate.edu!news.iastate.edu!rhawkins
Xref: chonsp.franklin.ch alt.folklore.computers:59256

In article <3960FD28.889A6941@netcologne.de>,
Werner Uelpenich  <werner.uelpenich@netcologne.de> wrote:

>In unix shells you call a lot of small utilities with cryptic short names, in
>DCL you call a function by a well memorizable name undependant from the
>taskname which does the work at last.

cryptic?  rm, vi, dd . . . . perfect sense.  Never trust a command
with more than two letters . . . :)

hawk, who has been meaning to file bug reports against mkdir and rmdir
for some time, as any command of such importance should not be defiled
with a long name . . .

-- 
Richard E. Hawkins
dochawk@psu.edu [regardless of where the message says it comes from]

These opinions will not be those of  Penn State until they pay my retainer.

######

Reply-To: "Xsemster" <abuse@aol.com>
From: "Xsemster" <abuse@aol.com>
Newsgroups: alt.folklore.computers
References: <873dlsdrci.fsf@cartman.azz.net> <877lb4f18g.fsf@litterbox.meowing.net> <cNT75.4138$7%3.326535@news.flash.net> <8k583b$mgh@teal.sni.net>
Subject: Re: Features to save from old OSs
Date: Fri, 7 Jul 2000 16:00:59 -0500
Organization: Monkey Spank
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.00.2615.200
X-MimeOLE: Produced By Microsoft MimeOLE V5.00.2615.200
Lines: 14
Message-ID: <3966453c$0$94738$726baab@news.execpc.com>
NNTP-Posting-Host: 3e0796fd.news.execpc.com
X-Trace: DXC=DQVg@6Eo1iOb0EN^dS;UdE`<WTGT>OiBNbCigiU>8_hH4`>8YbCYbkM9<ZNncHT^=D:\YZeYBnTHLT77B;ZJ6UmKU0H@G1m=7JJ
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!enews.sgi.com!newsxfer.eecs.umich.edu!news-spur1.maxwell.syr.edu!news.maxwell.syr.edu!newsfeed.skycache.com!Cidera!152.163.239.129!portc01.blue.aol.com!newsfeeds.sol.net!news.execpc.com!newspeer.sol.net!posts0.nwblwi.newsops.execpc.com!posts.news.execpc.com!not-for-mail
Xref: chonsp.franklin.ch alt.folklore.computers:59297

I know it was a toy, and I know it will probably start a flame war on some
distant world, but I can't help but feel nostalgic for the ol' Commodore 64
and its ancestors the PET 4016 and PET 8032. The PETs really defined what a
desktop PC should look like up until the "tower" look.

What was best about the OS, aside from being brainnumbingly simple, was
entering BASIC code-- almost any command or instruction could be entered
with just two key strokes-- LOAD could be shortened to L [SHIFT]O, SAVE to S
[SHIFT]A, etc.

Typing out Visual Anything is fine with all the pop-ups you can turn on, but
it still isn't as fast as the Commie 64 was.

######

From: gah@ugcs.caltech.edu (glen herrmannsfeldt)
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: 7 Jul 2000 21:58:17 GMT
Organization: California Institute of Technology, Pasadena
Lines: 30
Message-ID: <8k5jpp$dei@gap.cco.caltech.edu>
References: <873dlsdrci.fsf@cartman.azz.net> <8jrmif$2l0$1@agate.berkeley.edu>      <sit5msscksqbrr7o6kdetbd209t6267gsv@4ax.com> <8k1qvj$mph$1@bob.news.rcn.net> <rvc95.715$jE1.45272@typhoon.ne.mediaone.net>
NNTP-Posting-Host: seniti.ugcs.caltech.edu
X-Newsreader: NN version 6.5.0 #1 (NOV)
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news.maxwell.syr.edu!newsfeed.stanford.edu!uchinews!nntp-server.caltech.edu!gah
Xref: chonsp.franklin.ch alt.folklore.computers:59285

Ric Werme <werme@nospam.mediaone.net> writes:

(snip)
>Not really, the idea here is that if you have a big database, keep
>it on disk for safe-keeping, keep it in RAM for fast access.

>Celera Genomics has a DEC, err, Compaq Wildfire, err, GS160 system
>with 64 GB of RAM, which I think pretty much matches their library of
>human DNA fragments.  To assemble everything in the right order they
>have to compare a fragment with all the rest and look for matching
>stretches.  If they're lucky, a couple matches let them combine a few
>pieces together.

>It's a lot faster to do all this when the data is in RAM and stays there.
(snip)
>It was only a few years ago that big Alpha servers crossed the 4 GB
>boundary, that we're up to 64 GB means that we're eating up addressing
>bits at a rate well in excess of Moore's Law.  Of course, no one will
>ever need more than 64 bits of addressing space, right?

IBM made a machine 20 years ago with a 64 bit (virtual) address space.
I think it was System/38, though I could have the number wrong.

We are definitely out of 32 bits, though.  I remember when the i386
came out, and everyone was glad to have a huge flat 32 bit address space.
Well, it isn't enough.  There are very few compilers that can generate
segmented x86 code so we can have more than 3.5GB (Intel Solaris) user space.
Note that the pentium II/III have a 36 bit (real) address bus!

-- glen

######

Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
References: <873dlsdrci.fsf@cartman.azz.net> <FxArnK.401@spcuna.spc.edu> <962926887.864144@shelley.paradise.net.nz> <FxB09n.232@spcuna.spc.edu>
Organization: Daedalus Consulting
X-Newsreader: trn 4.0-test72 (19 April 1999)
From: don@news.daedalus.co.nz (Don Stokes)
Message-ID: <963063231.193206@shelley.paradise.net.nz>
Cache-Post-Path: shelley.paradise.net.nz!unknown@203-96-144-16.cable.paradise.net.nz
X-Cache: nntpcache 2.4.0b2 (see http://www.nntpcache.org/)
Lines: 24
Date: Sat, 08 Jul 2000 13:34:09 GMT
NNTP-Posting-Host: 203.96.152.26
X-Complaints-To: newsadmin@xtra.co.nz
X-Trace: news.xtra.co.nz 963063249 203.96.152.26 (Sun, 09 Jul 2000 01:34:09 NZST)
NNTP-Posting-Date: Sun, 09 Jul 2000 01:34:09 NZST
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!enews.sgi.com!news.xtra.co.nz!not-for-mail
Xref: chonsp.franklin.ch alt.folklore.computers:59311

Terry Kennedy  <terry@gate.tmk.com> wrote:
[DECnet-PLUS/DECnet/OSI/Phase 5 DECnet/whatever it's called this week]
>  When the product finally came out, people looked at it and barfed, most-
>ly due to the command interface and exceedingly verbose logging.

NCL killed DECnet/OSI.  NCP, once you get your head around the concepts
of volatile and permanent databases and the various object types you have
to deal with, is moderately intuitive.  NCL quite simply isn't.

>  Did it ever actually get to the point where someone could say it had
>"been"? I recall it more as "it's coming! it's coming! it's coming! <blink>
>what? never heard of it!"

<grin> You don't live in a capital city.  8-)  There was this thing
called GOSIP -- the Government Open Systems Interconnect Profile, in
which governments mandated the use of OSI when purchasing networking
products for government use.  Admittedly, this thing was observed mainly
in the breach, before the governments finally saw sense and admitted
that OSI wasn't going anywhere (the OSI lot still hasn't agreed on
little things like network terminals, ie what Telnet had been doing for
20 years) and that there was rapid convergence to TCP/IP which rendered
any attempts to specify protocols at government level irrelevant.

-- don

######

From: jmfbahciv@aol.com
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: Sun, 09 Jul 00 10:37:10 GMT
Organization: UltraNet Communications, Inc.
Lines: 24
Message-ID: <8k9utn$3kd$2@bob.news.rcn.net>
References: <873dlsdrci.fsf@cartman.azz.net> <877lb4f18g.fsf@litterbox.meowing.net> <cNT75.4138$7%3.326535@news.flash.net> <8k583b$mgh@teal.sni.net> <hnsngr-ya023180000807001837020001@news.pacbell.net> <m3g0pk2b7a.fsf@localhost.localdomain>
X-Trace: 5md8LqoLFJwm/hChbkqc4G4wNiQvprH3WwwfWUpEaTU=
X-Complaints-To: abuse@rcn.com
NNTP-Posting-Date: 9 Jul 2000 13:32:39 GMT
X-Newsreader:  News Xpress Version 1.0 Beta #4
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news.maxwell.syr.edu!feed1.news.rcn.net!rcn!216-164-247-91
Xref: chonsp.franklin.ch alt.folklore.computers:59422

In article <m3g0pk2b7a.fsf@localhost.localdomain>,
   Dowe Keller <dowe@localhost.localdomain> wrote:
>hnsngr@sirius.com (Ron Hunsinger) writes:
>
>> In article <8k583b$mgh@teal.sni.net>, bediger@csn.net (Bruce Ediger) 
wrote:
>
>> EBCDIC also, but I didn't mind that. Every once in a while I'd hear 
rumors
>> of this other character set called ASCII, but it seemed mighty strange 
to
>> me, with not even any gaps in the codes for the letters. How was I 
supposed
>> to map character codes to punched cards? Besides, didn't God use EBCDIC?
>                                                    ^^^^^^^^^^^^^^^^^^^^^^
>Actually, that would explain an awful lot of the seeming randomness in the
>universe :-)
>

And the Devil's in the details.

/BAH

Subtract a hundred and four for e-mail.

######

From: andrew@cucumber.demon.co.uk (Andrew Gabriel)
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: 8 Jul 2000 10:56:23 GMT
Organization: home
Message-ID: <8k71cn$2a9@cucumber.demon.co.uk>
References: <873dlsdrci.fsf@cartman.azz.net>
NNTP-Posting-Host: cucumber
X-NNTP-Posting-Host: cucumber.demon.co.uk:158.152.58.86
X-Trace: news.demon.co.uk 963084165 nnrp-09:27569 NO-IDENT cucumber.demon.co.uk:158.152.58.86
X-Complaints-To: abuse@demon.net
X-Newsreader: knews 0.9.6
Lines: 23
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news.maxwell.syr.edu!dispose.news.demon.net!demon!news.demon.co.uk!demon!cucumber.demon.co.uk!usenet
Xref: chonsp.franklin.ch alt.folklore.computers:59448

In article <873dlsdrci.fsf@cartman.azz.net>,
	Adam Sampson <azz@gnu.org> writes:

Kernel modules and device drivers to operate in separate protected
environments which:
a) Don't lose whole system if they crash (at least, configurable per module)
b) Debug in exactly same way as user-level processes
c) Can be reloaded if necessary (e.g. after crash or to try new version).

GEC OS4000 did this. Actually, pretty much everything a unix kernel does
was run as a process with full protection wrapped round it. Anything which
couldn't run as a process was implemented in hardware (e.g. the inter-
process cummunication, short term scheduling, protection, error reporting).
There was no privileged kernel/supervisor code.

All i/o asynchronous. Synchronous i/o provided by means of library calls,
but system architected for asynchronous i/o throughout rather than as
patched-in after thoughts done differently in different places.

-- 
Andrew Gabriel
Consultant Software Engineer

######

From: Brian Inglis <Brian.dot.Inglis@SystematicSw.ab.ca>
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: Sat, 08 Jul 2000 15:04:26 -0600
Organization: Systematic Software
Reply-To: Brian.dot.Inglis@SystematicSw.ab.ca
Message-ID: <29uems4ns48taimsccgsrnnmqhlnblklce@4ax.com>
References: <873dlsdrci.fsf@cartman.azz.net> <8jrmif$2l0$1@agate.berkeley.edu> <sit5msscksqbrr7o6kdetbd209t6267gsv@4ax.com> <8k286i$jgh$1@agate.berkeley.edu>
X-Newsreader: Forte Agent 1.8/32.548
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
NNTP-Posting-Host: 207.148.136.135
X-Original-NNTP-Posting-Host: 207.148.136.135
X-Trace: 8 Jul 2000 15:04:28 -0700, 207.148.136.135
Lines: 165
X-Original-NNTP-Posting-Host: 204.50.1.43
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!nyc-news-feed1.bbnplanet.com!news.gtei.net!colt.net!news.maxwell.syr.edu!howland.erols.net!news.idt.net!nntp.cadvision.com!news.cadvision.com!207.148.136.135
Xref: chonsp.franklin.ch alt.folklore.computers:59402

On 6 Jul 2000 15:21:54 GMT, korpela@sagan.ssl.berkeley.edu (Eric
J. Korpela) wrote:

>In article <sit5msscksqbrr7o6kdetbd209t6267gsv@4ax.com>,
>Brian Inglis  <Brian.dot.Inglis@SystematicSw.ab.ca> wrote:
>>On 4 Jul 2000 03:44:15 GMT, korpela@ellie.ssl.berkeley.edu (Eric
>>J. Korpela) wrote:
>>
>>>In article <873dlsdrci.fsf@cartman.azz.net>, Adam Sampson  <azz@gnu.org> wrote:
>>>>                             block-orientated IO for database
>>>>operations
>>>
>>>Sheesh, this is the new century and we're still doing explicit I/O to 
>>>storage devices?  Either the memory you're accessing is backed by permanent
>>>storage or its not.  The only difficulty is when you need to exceed your
>>>address space.  Block oriented explicit I/O can be written on top of this
>>>if necessary for legacy application support.

Explicit I/O to devices has the advantage that it can be managed
to provide the desired result: fast application I/O regardless of
the effect on other applications/subsystems. 

Few systems have attempted to optimize virtual memory subsystems
to perform efficient I/O; most just read/write a block/page at a
time. 

Addressing 2^64/16x10^18 bytes is a limit for which apps that
don't try to model the universe in detail? Any systems with that
much disk space/backing store? 

>>>What databases need is not block-oriented I/O, but a means of serializing
>>>memory access across multiple processes running on multiple systems.  The
>>
>>Ugh -- shades of VMS clustering -- slowest I/O known to man --
>>databases need fast I/O to succeed -- serializing updates across
>>multiple systems uses two phase commit -- with explicit
>>recognition that it's faster if it's all on one system. 
>
>Which is why it's optional.  You don't do all I/O as serialized, you only
>do specifically requested page swaps as serialized.  And there are times 
>you can't do it all on one machine.  If you think otherwise you haven't 
>done anything large.

If you think in terms of single blocks/pages, or over networks,
you haven't done anything fast. 

>>>serialized/non-serialized writes and automatic update.  The system for which 
>>>the storage is local needs to handle the serialization.  That means it needs 
>>
>>All memory should be local to the system, otherwise why pay the
>>high price for fast access, and why bother with multiple systems
>>(clusters) when you can just hang a bunch of engines, GBs of
>>memory, and TBs of disk on one system, and use hierarchical MP
>>locks within a single system. 
>
>1. It's cheaper.
>2. Even in expensive systems there are only so many CPUs and
>only so much memory you can add.  What happens when 8 or 16 or
>256 CPUs are no longer enough?

Cheaper isn't going to be faster: clustering approaches give a
low performance level for a low price; buy a bigger system with
more memory, more and faster CPUs, more controllers and drives,
to get higher performance. 

Eliminate the slow operations, overhead and bottlenecks; balance
the work across all the functional units available. If you have a
CPU horsepower problem, do the work differently, optimize
existing algorithms or code, find better algorithms, change the
approach to make something else the bottleneck that you can deal
with by expansion. 

>>>to be in the network protocol that handles I/O to storage devices.  For 
>>
>>Obviously no need for performance if you're doing I/O over
>>network protocols -- why not just use magtapes (and optional
>>station wagons -- or minivans now?) 
>
>Haven't noticed gigabit ethernet, have you?  Networks are getting
>faster in case you haven't noticed.

Hopefully also, allowing much larger packet sizes (MB) and
avoiding CDMA to allow sustained load > 20%, in which case, it
probably should not be termed enet. Not much use having speed if
you can't use all of it all the time. 

>>>example, if a system has mapped serialized a page for which the physical 
>>>storage is managed by another system and the page is in local volatile storage,
>>
>>Why bother wasting memory if the storage is on another system
>>across a network -- shared drives would be faster. 
>
>What the hell does it matter what type of network it is?  Or don't
>you think shared drives constitute a network?  Do you think there's
>a need to arbitrate accesses to a shared drive?

Why move a page over a network, if it's faster to write it (and
hopefully a large bunch of its friends) to a shared drive and
send a packet (or better, update shared memory) to say what to
read and where. 

>>>the local page will be invalidated by a memory write to the same page by
>>>any other system.  For those familiar with unix mmap(), it means that
>>>there should be another mapping mode MAP_SERIAL, which is the equivalent of
>>>MAP_SHARED working network wide with automatic access control semaphores.
>>
>>Or just use cross system shared file locking if you have such
>>meagre requirements. 
>
>In case you haven't noticed, it's a database.  You certainly don't want
>to lock files, you want to lock the smallest addressable portion, i.e. the
>page.  You do it in the VM system.  You write your explicit I/O to work
>through the VM system.

I would never dream of locking a file or a table, that could
cause an unacceptable level of contention; I assume the semantics
would allow record/row level logical resource locks; even
block/page level physical locks are too large; look at Sybase DBs
for the problems caused and workarounds required to alleviate the
situation. 

>>Some of use like to see our I/O move at high MB/s rates rather
>>than low Mb/s rates. Like using Alpha servers with lots of
>>engines and big memory -- to avoid the I/O as much as possible. 
>
>Of course, that's why you memory map files.  But what happens when your
>alpha server runs out of oomph?  There are only so many CPU slots.

Look at what is causing the bottleneck and eliminate it or move
it elsewhere, where you can do something about it. 

>How do prevent losing data in your database when the system crashes
>if you aren't doing I/O?  

Thrashing or excessive waits or deadlocked or system hung? 

>How do you avoid I/O when your database is 15 terabytes in size?  

Add more memory (see Rick Werme's post earlier in thread) and
more arms (disk farms) connected over multiple high speed
channels to multiple controllers to multiple CPUs. You speed up
I/O by: not doing it (database caching); doing it faster (raw
async device I/O, non-volatile write through cache on disk
controllers, elevator device scheduling in the controller);
eliminating contention and waits (multi-path I/O dispatching,
independent path I/O completion, balancing I/O loads across paths
and devices, avoiding hot spots!) 

Proper configuration planning, regular reevaluation of what the
system is actually doing at peak times, and actions to eliminate
the slowdowns that are occurring work wonders. 

I believe that most management nowadays would rather go to lunch
with salespeople and be sold a system that promises to make their
problems go away, than listen to analysts who can make their
systems do much more, at the cost of some manpower and
application downtime; they avoid mentioning that migrating to new
systems often involves massive disruption to users ability to
perform useful work. 

>Eric
Thanks. Take care, Brian Inglis 	Calgary, Alberta, Canada
-- 
Brian_Inglis@CSi.com 	(Brian dot Inglis at SystematicSw dot ab dot ca)
				use address above to reply

######

From: korpela@ellie.ssl.berkeley.edu (Eric J. Korpela)
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: 8 Jul 2000 23:34:33 GMT
Organization: Cal Berkeley-- Space Sciences Lab
Lines: 131
Message-ID: <8k8dq9$62b$1@agate.berkeley.edu>
References: <873dlsdrci.fsf@cartman.azz.net> <sit5msscksqbrr7o6kdetbd209t6267gsv@4ax.com> <8k286i$jgh$1@agate.berkeley.edu> <29uems4ns48taimsccgsrnnmqhlnblklce@4ax.com>
NNTP-Posting-Host: ellie.ssl.berkeley.edu
X-Trace: agate.berkeley.edu 963099273 6219 128.32.18.173 (8 Jul 2000 23:34:33 GMT)
X-Complaints-To: abuse@berkeley.edu
NNTP-Posting-Date: 8 Jul 2000 23:34:33 GMT
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!nyc-news-feed1.bbnplanet.com!news.gtei.net!colt.net!news.maxwell.syr.edu!news-hog.berkeley.edu!ucberkeley!agate.berkeley.edu!agate!ellie.ssl.berkeley.edu!korpela
Xref: chonsp.franklin.ch alt.folklore.computers:59438

In article <29uems4ns48taimsccgsrnnmqhlnblklce@4ax.com>,
Brian Inglis  <Brian.dot.Inglis@SystematicSw.ab.ca> wrote:
>>>>                  Either the memory you're accessing is backed by permanent
>>>>storage or its not.  The only difficulty is when you need to exceed your
>>>>address space.  Block oriented explicit I/O can be written on top of this
>>>>if necessary for legacy application support.
>
>Explicit I/O to devices has the advantage that it can be managed
>to provide the desired result: fast application I/O regardless of
>the effect on other applications/subsystems. 

Explicit I/O is rarely explicit, but usually has to pass through cache
subsystems.  Unifying the I/O and VM subsystems usually makes sense.

>Few systems have attempted to optimize virtual memory subsystems
>to perform efficient I/O; most just read/write a block/page at a
>time. 

Sorry to disappoint, but I've seen very few explicit I/O subsystems that
don't read/write a block/page at a time.  Since the most common type of
I/O is a page fault, that should probably be the most optimized bit of 
I/O code.  Since every form of block I/O should be able to be performed 
through a decently designed VM subsystem, there's no need to separate them.  
The only exception would be devices that can only be accessed sequentially. 

>>Which is why it's optional.  You don't do all I/O as serialized, you only
>>do specifically requested page swaps as serialized.  And there are times 
>>you can't do it all on one machine.  If you think otherwise you haven't 
>>done anything large.
>
>If you think in terms of single blocks/pages, or over networks,
>you haven't done anything fast. 

Given that we're talking about random database record access, single 
blocks/pages are about the best you're going to be able to do.  And
when access over a network is necessary, there needs to be a synchronization
mechanism.

>>1. It's cheaper.
>>2. Even in expensive systems there are only so many CPUs and
>>only so much memory you can add.  What happens when 8 or 16 or
>>256 CPUs are no longer enough?
>
>Cheaper isn't going to be faster: clustering approaches give a
>low performance level for a low price; buy a bigger system with
>more memory, more and faster CPUs, more controllers and drives,
>to get higher performance. 

Well, you certainly can't go any faster than you can afford.  And
right now, a database can be larger than the memory of the largest
system you can afford.  You may require more CPUs than you can put 
in a single system.  Distributed databases are, and will continue to
be a reality.  

>Eliminate the slow operations, overhead and bottlenecks; balance
>the work across all the functional units available.

Those functional units may, by necessity be on different, widely
separated systems.

>If you have a CPU horsepower problem, do the work differently, optimize
>existing algorithms or code, find better algorithms, change the
>approach to make something else the bottleneck that you can deal
>with by expansion. 

There are some problems that can't be optimized away.

>>What the hell does it matter what type of network it is?  Or don't
>>you think shared drives constitute a network?  Do you think there's
>>a need to arbitrate accesses to a shared drive?
>
>Why move a page over a network, if it's faster to write it (and
>hopefully a large bunch of its friends) to a shared drive and
>send a packet (or better, update shared memory) to say what to
>read and where. 

Hearing problems?  Regardless of the type of connection, a shared drive
IS A NETWORK.  Locking and serialization are still a problem.

>>>>the local page will be invalidated by a memory write to the same page by
>>>>any other system.  For those familiar with unix mmap(), it means that
>>>>there should be another mapping mode MAP_SERIAL, which is the equivalent of
>>>>MAP_SHARED working network wide with automatic access control semaphores.
>>>
>>>Or just use cross system shared file locking if you have such
>>>meagre requirements. 
>>
>>In case you haven't noticed, it's a database.  You certainly don't want
>>to lock files, you want to lock the smallest addressable portion, i.e. the
>>page.  You do it in the VM system.  You write your explicit I/O to work
>>through the VM system.
>
>I would never dream of locking a file or a table, that could
>cause an unacceptable level of contention; I assume the semantics
>would allow record/row level logical resource locks; even
>block/page level physical locks are too large; look at Sybase DBs
>for the problems caused and workarounds required to alleviate the
>situation. 

I'm certainly not saying page locks are a complete solution.  They are
a big part of the solution.

>>Of course, that's why you memory map files.  But what happens when your
>>alpha server runs out of oomph?  There are only so many CPU slots.
>
>Look at what is causing the bottleneck and eliminate it or move
>it elsewhere, where you can do something about it. 

That's assuming there is somewhere you can move it and that there is
something you can do about it.

>>How do you avoid I/O when your database is 15 terabytes in size?  
>
>Add more memory (see Rick Werme's post earlier in thread) and
>more arms (disk farms) connected over multiple high speed
>channels to multiple controllers to multiple CPUs. 

In other words, you use a network. 

>Proper configuration planning, regular reevaluation of what the
>system is actually doing at peak times, and actions to eliminate
>the slowdowns that are occurring work wonders. 

Yes, and having support for some of this in the OS might help.

Eric
-- 
Eric Korpela                        |  An object at rest can never be
korpela@ssl.berkeley.edu            |  stopped.
<a href="http://sag-www.ssl.berkeley.edu/~korpela">Click for home page.</a>

######

From: hnsngr@sirius.com (Ron Hunsinger)
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Message-ID: <hnsngr-ya023180000807001837020001@news.pacbell.net>
References: <873dlsdrci.fsf@cartman.azz.net> <877lb4f18g.fsf@litterbox.meowing.net> <cNT75.4138$7%3.326535@news.flash.net> <8k583b$mgh@teal.sni.net>
Organization: ErsteSoft
Mime-Version: 1.0
Content-Type: text/plain; charset=ISO-8859-1
Content-transfer-encoding: 8bit
X-Newsreader: Yet Another NewsWatcher 2.3.1
Lines: 19
Date: Sat, 08 Jul 2000 18:37:02 -0700
NNTP-Posting-Host: 216.103.86.8
X-Complaints-To: abuse@pacbell.net
X-Trace: news.pacbell.net 963106550 216.103.86.8 (Sat, 08 Jul 2000 18:35:50 PDT)
NNTP-Posting-Date: Sat, 08 Jul 2000 18:35:50 PDT
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!News.Amsterdam.UnisourceCS!newshunter!cosy.sbg.ac.at!newsfeed.Austria.EU.net!nmaster.kpnqwest.net!npeer.kpnqwest.net!nntp.primenet.com!nntp.gblx.net!news-out.nibble.net!news-in.nibble.net!cyclone-sf.pbi.net!206.13.28.144!news.pacbell.net.POSTED!hnsngr
Xref: chonsp.franklin.ch alt.folklore.computers:59481

In article <8k583b$mgh@teal.sni.net>, bediger@csn.net (Bruce Ediger) wrote:

> The misdesigns in NOS included single-level directory structure,
> limited length file names (10 character?), non-ASCII character set, 
> and bizarrely over-complicated I/O scheme(s).

You had 10 characters? I would have killed for 7.

Burroughs Medium Systems (er, excuse me, they're now called "Unisys
V-Series", but this being AFC I can confess that I still think of them
under the old name) had (still has!) a single-level directory structure
with 6-character file names.

EBCDIC also, but I didn't mind that. Every once in a while I'd hear rumors
of this other character set called ASCII, but it seemed mighty strange to
me, with not even any gaps in the codes for the letters. How was I supposed
to map character codes to punched cards? Besides, didn't God use EBCDIC?

-Ron Hunsinger

######

Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
References: <873dlsdrci.fsf@cartman.azz.net> <877lb4f18g.fsf@litterbox.meowing.net> <cNT75.4138$7%3.326535@news.flash.net> <8k583b$mgh@teal.sni.net> <hnsngr-ya023180000807001837020001@news.pacbell.net>
From: Dowe Keller <dowe@localhost.localdomain>
Date: 08 Jul 2000 21:12:57 -0700
Message-ID: <m3g0pk2b7a.fsf@localhost.localdomain>
Lines: 17
X-Newsreader: Gnus v5.5/Emacs 20.3
NNTP-Posting-Host: 20923419669.sierratel.com
X-Original-NNTP-Posting-Host: 20923419669.sierratel.com
X-Trace: 8 Jul 2000 21:32:38 -0700, 20923419669.sierratel.com
Organization: news.sierratel.com
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!enews.sgi.com!newsxfer.eecs.umich.edu!news-spur1.maxwell.syr.edu!news.maxwell.syr.edu!newsfeed.skycache.com!Cidera!209.155.26.10!news.sierratel.com!20923419669.sierratel.com
Xref: chonsp.franklin.ch alt.folklore.computers:59476

hnsngr@sirius.com (Ron Hunsinger) writes:

> In article <8k583b$mgh@teal.sni.net>, bediger@csn.net (Bruce Ediger) wrote:

> EBCDIC also, but I didn't mind that. Every once in a while I'd hear rumors
> of this other character set called ASCII, but it seemed mighty strange to
> me, with not even any gaps in the codes for the letters. How was I supposed
> to map character codes to punched cards? Besides, didn't God use EBCDIC?
                                                    ^^^^^^^^^^^^^^^^^^^^^^
Actually, that would explain an awful lot of the seeming randomness in the
universe :-)

-- 
dowe@sierratel.com
This is the theory that Jack built.
This is the flaw that lay in the theory that Jack built.
This is the palpable verbal haze that hid the flaw that lay in...

######

From: jmfbahciv@aol.com
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: Tue, 11 Jul 00 09:35:39 GMT
Organization: UltraNet Communications, Inc.
Lines: 44
Message-ID: <8kf0ih$9sm$1@bob.news.rcn.net>
References: <873dlsdrci.fsf@cartman.azz.net> <025a9e75.2d4e185c@usw-ex0108-063.remarq.com> <87g0pmg9e8.fsf@cartman.azz.net> <8k71of$qdd$4@bob.news.rcn.net> <8kd5ab$4vr$1@news.iastate.edu>
X-Trace: IIXfmWlgFbadWrvuhY8eBXviusFJDsCr2u2KnBXd6YU=
X-Complaints-To: abuse@rcn.com
NNTP-Posting-Date: 11 Jul 2000 11:31:29 GMT
X-Newsreader:  News Xpress Version 1.0 Beta #4
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news.maxwell.syr.edu!feed1.news.rcn.net!rcn!207-172-245-214
Xref: chonsp.franklin.ch alt.folklore.computers:59531

In article <8kd5ab$4vr$1@news.iastate.edu>,
   rhawkins@iastate.edu (Rick Hawkins) wrote:
>In article <8k71of$qdd$4@bob.news.rcn.net>,  <jmfbahciv@aol.com> wrote:
>>In article <87g0pmg9e8.fsf@cartman.azz.net>,
>>   Adam Sampson <azz@gnu.org> wrote:
>
>>>Actually, Linux is currently implementing this, by way of adding a
>>>list of capabilites (Linux's implementation of privileges) to ELF
>>>executables. Presumably you need the setuid bit set to actually make
>>>it follow the suggested capabilities, and it means that you can't use
>>>the same trick with scripts.
>
>>>It seems to be agreed that this is only to be used as a kludge until
>>>the filesystem can store the capability set...
>
>>But you don't have to store it.  You can have the user define
>                       ^^^^^
>>access via a file.  The only change to the filesystem is
>>agreement on when the protection extension capability gets called.
>
>OK, I'll bite.  Just what kind of filestystem doesn't store things? :)

Oh, there are lots of them, usually written by people who have
small computer thinking.  

There's a difference in hard-coding protection schemes and
allowing a program to figure out protection schemes.  TOPS-10
started out by having hard-coded protection schemes on a
per-file and a per-UFD basis.  When JMF wrote the File Daemon,
he changed a hardcoded protection bit such that a program
was called to determine flavors of access.  You can get some
pretty powerful protection schemes that way.

>
>hawk, probably misreading this all, and deliberately avoiding the 
>obvious windows slam
>
I'd never slam the windows on you, Hawk.  I'd be afraid that
I'd miss a really good line of humor :-).  Glad to see you
back on-line...hope it lasts.

/BAH

Subtract a hundred and four for e-mail.

######

Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
References: <873dlsdrci.fsf@cartman.azz.net> <8jrmif$2l0$1@agate.berkeley.edu>
     <sit5msscksqbrr7o6kdetbd209t6267gsv@4ax.com> <8k1qvj$mph$1@bob.news.rcn.net>
     <rvc95.715$jE1.45272@typhoon.ne.mediaone.net> <8k4gku$1ji$5@bob.news.rcn.net>
From: Ric Werme <werme@nospam.mediaone.net>
X-Newsreader: NN version 6.5.0 CURRENT #119
Lines: 75
Message-ID: <tz9a5.2080$jE1.95237@typhoon.ne.mediaone.net>
Date: Mon, 10 Jul 2000 01:22:33 GMT
NNTP-Posting-Host: 24.91.12.15
X-Complaints-To: abuse@mediaone.net
X-Trace: typhoon.ne.mediaone.net 963192153 24.91.12.15 (Sun, 09 Jul 2000 21:22:33 EDT)
NNTP-Posting-Date: Sun, 09 Jul 2000 21:22:33 EDT
Organization: Road Runner
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!newsfeeds.belnet.be!news.belnet.be!news-spur1.maxwell.syr.edu!news.maxwell.syr.edu!feed1.news.rcn.net!rcn!chnws02.mediaone.net!chnws05.ne.mediaone.net!24.128.8.70!typhoon.ne.mediaone.net.POSTED!not-for-mail
Xref: chonsp.franklin.ch alt.folklore.computers:59604

jmfbahciv@aol.com writes:

>In article <rvc95.715$jE1.45272@typhoon.ne.mediaone.net>,
>   Ric Werme <werme@nospam.mediaone.net> wrote:
>>jmfbahciv@aol.com writes:

>>Not really, the idea here is that if you have a big database, keep
>>it on disk for safe-keeping, keep it in RAM for fast access.

>That implies that there are data bases where all elements are
>getting modified often.  I thought most databases consisted
>of data that was mostly static.

Not at all - there are lots of applications where complicated searches
or data mining wind up looking a lot of the data.  If the "working set"
doesn't fit in RAM, you need more RAM.

>>Celera Genomics has a DEC, err, Compaq Wildfire, err, GS160 system

>That renaming of products is going to kill them faster than their
>inability to make them.  You might pass that comment on.

The DEC, err, Compaq part is water over the dam.  The Wildfire, err, GSxxx
part is standard let the code name leak to the trade press, compounded with
a boring official name and infringement on the "Wildfire" trademark held
by someone else.  The bigger problem was the delays in getting the
machine out.  The company has done a much better than usual job of getting
the story out to customers and training sales folk.

>>It's a lot faster to do all this when the data is in RAM and stays there.
>>So, fire up the assembly application, read umpteen GB of raw DNA
>>sequences and start matching.

>Right.  I guess people are using the term "data base" in a looser
>way than when I was out there :-).  There was data base and
>then there was computational.  

There's also speed, witness the TPC benchmark. See
http://www.tpc.org/new_result/ttperf.idc.  IBM just blew everyone out
of the water with $14,000,000 of clustered Windows 2000 systems.  Just
running these benchmarks are a huge investment.  For high end systems
it takes weeks of system time to build the databases to be used in
running the benchmark!  I think that IBM run used some 2,000 big
disks, so I guess this is a case where the only a subset is in RAM,
most of the data is on disk.

>>The code is tiny, the OS is tiny.  Compile for speed, don't worry
>>about code size.

>Well, that's an attitude that I haven't understood.  (I used to
>blame it on those COBOL programmers that didn't know any better.)

I assume you mean the "don't worry about code" size part.

If a key loop in the code is 4X bigger due to loop unrolling or other
speed optimizations, that's okay.  If it no longer fits in the
instruction cache, that's very bad.  Database coders know that or
they're out of business.  Your Cobol programmers didn't know that
and that was mostly okay for the tasks at hand.

The same applies to weather and climatology forecasting, protein
folding simulations, circuit emulation (emulating the latest Alpha
CPUS requires machines with more than 32 bits of addressing) Multi GB
data sets, programs from 32 KB to 25 MB?.  Code tiny, data big.  Buy
enough RAM to hold the data you need.

Back when memory was a buck a byte, you bought what you could afford
and worked around the problem.  Today, the cost of a programmer to
keep code small would buy far more RAM than another programmer could
fill with the worst code imaginable.

	-Ric
--
Ric Werme                            | werme@nospam.mediaone.net
http://people.ne.mediaone.net/werme  |       ^^^^^^^ delete

######

From: "P Linnane" <plyons@ca-unlimited.com>
Newsgroups: alt.folklore.computers
References: <873dlsdrci.fsf@cartman.azz.net> <8jrmif$2l0$1@agate.berkeley.edu>    <sit5msscksqbrr7o6kdetbd209t6267gsv@4ax.com> <8k1qvj$mph$1@bob.news.rcn.net>    <rvc95.715$jE1.45272@typhoon.ne.mediaone.net> <8k4gku$1ji$5@bob.news.rcn.net> <tz9a5.2080$jE1.95237@typhoon.ne.mediaone.net>
Subject: Re: Features to save from old OSs
Lines: 67
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.00.2919.6600
X-MimeOLE: Produced By Microsoft MimeOLE V5.00.2919.6600
Organization: Ye 'Ol Disorganized NNTPCache groupie
Message-ID: <963206341.272445@server16.cable.com>
Cache-Post-Path: server16.cable.com!unknown@vp1-11.xcessible.com
X-Cache: nntpcache 2.4.0b2 (see http://www.nntpcache.org/)
Date: Mon, 10 Jul 2000 01:15:34 -0400
NNTP-Posting-Host: 204.92.64.17
X-Trace: nnrp1.uunet.ca 963206132 204.92.64.17 (Mon, 10 Jul 2000 01:15:32 EDT)
NNTP-Posting-Date: Mon, 10 Jul 2000 01:15:32 EDT
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!isdnet!newsfeed.cwix.com!torn!news.uunet.ca!nnrp1.uunet.ca.POSTED!not-for-mail
Xref: chonsp.franklin.ch alt.folklore.computers:59540

Hi Ric,

"Ric Werme" <werme@nospam.mediaone.net> wrote in message
news:tz9a5.2080$jE1.95237@typhoon.ne.mediaone.net...
> jmfbahciv@aol.com writes:
>
< snip snip >
>
> There's also speed, witness the TPC benchmark. See
> http://www.tpc.org/new_result/ttperf.idc.  IBM just blew everyone out
> of the water with $14,000,000 of clustered Windows 2000 systems.  Just
> running these benchmarks are a huge investment.  For high end systems
> it takes weeks of system time to build the databases to be used in
> running the benchmark!  I think that IBM run used some 2,000 big
> disks, so I guess this is a case where the only a subset is in RAM,
> most of the data is on disk.
>
> >>The code is tiny, the OS is tiny.  Compile for speed, don't worry
> >>about code size.
>
> >Well, that's an attitude that I haven't understood.  (I used to
> >blame it on those COBOL programmers that didn't know any better.)

(Hark, me ears is a burnin'). Generally speaking, a COBOL programmer would
optimize for small source code size ie: a small loop with many iterations or
seemingly (to some) superfluous PERFORM statements for reasons of improved
readability and future maintenance needs; something which was of less
priority to an Assembler or ForTran programmer.

>
> I assume you mean the "don't worry about code" size part.
>
> If a key loop in the code is 4X bigger due to loop unrolling or other
> speed optimizations, that's okay.  If it no longer fits in the
> instruction cache, that's very bad.  Database coders know that or
> they're out of business.  Your Cobol programmers didn't know that
> and that was mostly okay for the tasks at hand.
>
Of course <grin> there was that too. Though me'n'the boys still sequence the
source code with paging in mind.

How big was the instruction cache ?
>
> The same applies to weather and climatology forecasting, protein
> folding simulations, circuit emulation (emulating the latest Alpha
> CPUS requires machines with more than 32 bits of addressing) Multi GB
> data sets, programs from 32 KB to 25 MB?.  Code tiny, data big.  Buy
> enough RAM to hold the data you need.
>
Off-topic in any creative sense, but: how the **** do PC programs fill up
200MB on a disk ?
No, seriously, does anyone know ?
>
> Back when memory was a buck a byte, you bought what you could afford
> and worked around the problem.  Today, the cost of a programmer to
> keep code small would buy far more RAM than another programmer could
> fill with the worst code imaginable.
>
I'm afraid I'm gonna have to disagree with that sweeping statement. The cost
of processor-cache RAM isn't cheap. I'm currently phrasing an e-mail to the
effect that a macro-based 10 minute database merge/calc/print-to-file could
be brought down to 5-10 seconds using a compiler (and a programmer, of
course).

Rick

######

From: rhawkins@iastate.edu (Rick Hawkins)
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: 10 Jul 2000 18:40:11 GMT
Organization: Unknown
Lines: 28
Message-ID: <8kd5ab$4vr$1@news.iastate.edu>
References: <873dlsdrci.fsf@cartman.azz.net> <025a9e75.2d4e185c@usw-ex0108-063.remarq.com> <87g0pmg9e8.fsf@cartman.azz.net> <8k71of$qdd$4@bob.news.rcn.net>
NNTP-Posting-Host: pv2079.vincent.iastate.edu
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!enews.sgi.com!feeder.via.net!logbridge.uoregon.edu!newsrelay.iastate.edu!news.iastate.edu!rhawkins
Xref: chonsp.franklin.ch alt.folklore.computers:59544

In article <8k71of$qdd$4@bob.news.rcn.net>,  <jmfbahciv@aol.com> wrote:
>In article <87g0pmg9e8.fsf@cartman.azz.net>,
>   Adam Sampson <azz@gnu.org> wrote:

>>Actually, Linux is currently implementing this, by way of adding a
>>list of capabilites (Linux's implementation of privileges) to ELF
>>executables. Presumably you need the setuid bit set to actually make
>>it follow the suggested capabilities, and it means that you can't use
>>the same trick with scripts.

>>It seems to be agreed that this is only to be used as a kludge until
>>the filesystem can store the capability set...

>But you don't have to store it.  You can have the user define
                       ^^^^^
>access via a file.  The only change to the filesystem is
>agreement on when the protection extension capability gets called.

OK, I'll bite.  Just what kind of filestystem doesn't store things? :)

hawk, probably misreading this all, and deliberately avoiding the 
obvious windows slam

-- 
Richard E. Hawkins
dochawk@psu.edu [regardless of where the message says it comes from]

These opinions will not be those of  Penn State until they pay my retainer.

######

From: rhawkins@iastate.edu (Rick Hawkins)
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: 10 Jul 2000 18:43:29 GMT
Organization: Unknown
Lines: 21
Message-ID: <8kd5gh$45m$1@news.iastate.edu>
References: <873dlsdrci.fsf@cartman.azz.net> <8k71cn$2a9@cucumber.demon.co.uk>
NNTP-Posting-Host: pv2079.vincent.iastate.edu
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!enews.sgi.com!nntp.flash.net!howland.erols.net!vixen.cso.uiuc.edu!newsrelay.iastate.edu!news.iastate.edu!rhawkins
Xref: chonsp.franklin.ch alt.folklore.computers:59528

In article <8k71cn$2a9@cucumber.demon.co.uk>,
Andrew Gabriel <andrew@cucumber.demon.co.uk> wrote:
>In article <873dlsdrci.fsf@cartman.azz.net>,
>       Adam Sampson <azz@gnu.org> writes:
>
>Kernel modules and device drivers to operate in separate protected
>environments which:
>a) Don't lose whole system if they crash (at least, configurable per module)


So a little flag to set, $CRASH_REST_OF_SYSTEM_WHILE_AT_IT

:)

hawk

-- 
Richard E. Hawkins
dochawk@psu.edu [regardless of where the message says it comes from]

These opinions will not be those of  Penn State until they pay my retainer.

######

From: bhk@dsl.co.uk (Brian {Hamilton Kelly})
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: Mon, 10 Jul 2000 20:05:10 GMT
Organization: Dragonhill Systems Ltd
Message-ID: <963259510snz@dsl.co.uk>
References: <873dlsdrci.fsf@cartman.azz.net> <8jrmif$2l0$1@agate.berkeley.edu>
X-Trace: mail2news.demon.co.uk 963262275 mail2news:12589 mail2news mail2news.demon.co.uk
X-Complaints-To: abuse@demon.net
X-Mail2News-Path: news.demon.net!dsl.demon.co.uk
X-Newsreader: Demon Internet Simple News v1.31
Lines: 14
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!enews.sgi.com!newsxfer.eecs.umich.edu!news-spur1.maxwell.syr.edu!news.maxwell.syr.edu!dispose.news.demon.net!demon!news.demon.co.uk!demon!mail2news.demon.co.uk!not-for-mail
Xref: chonsp.franklin.ch alt.folklore.computers:59617

In article <8jrmif$2l0$1@agate.berkeley.edu>
           korpela@ellie.ssl.berkeley.edu "Eric J. Korpela" writes:

> Sheesh, this is the new century and we're still doing explicit I/O to 
> storage devices?  

No it isn't; there's another 174 days until the start of the 21st Century.

-- 
Brian {Hamilton Kelly}                                          bhk@dsl.co.uk
    "We have gone from a world of concentrated knowledge and wisdom to one of
    distributed ignorance.  And we know and understand less while being incr-
    easingly capable."                          Prof. Peter Cochrane, BT Labs

######

From: bhk@dsl.co.uk (Brian {Hamilton Kelly})
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: Mon, 10 Jul 2000 20:07:30 GMT
Organization: Dragonhill Systems Ltd
Message-ID: <963259650snz@dsl.co.uk>
References: <873dlsdrci.fsf@cartman.azz.net> <025a9e75.2d4e185c@usw-ex0108-063.remarq.com> <396477A1.A772FA58@tnglwood.demon.co.uk>
X-Trace: mail2news.demon.co.uk 963262280 mail2news:12593 mail2news mail2news.demon.co.uk
X-Complaints-To: abuse@demon.net
X-Mail2News-Path: news.demon.net!dsl.demon.co.uk
X-Newsreader: Demon Internet Simple News v1.31
Lines: 29
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!enews.sgi.com!newsxfer.eecs.umich.edu!news-spur1.maxwell.syr.edu!news.maxwell.syr.edu!newsfeed.icl.net!dispose.news.demon.net!demon!news.demon.co.uk!demon!mail2news.demon.co.uk!not-for-mail
Xref: chonsp.franklin.ch alt.folklore.computers:59619

In article <396477A1.A772FA58@tnglwood.demon.co.uk>
           unclebob@tnglwood.demon.co.uk "AKA Uncle Bob" writes:

> martin2305 wrote:
> 
> > The ability for an executable image to possess file access
> > privileges *in its own right* (I don't think UNIX OSs have this -
> 
>  ISTR Maurice Wilkes telling me that the Titan operating system had
> this.

VMS executables could readily be given any of the 40+ privileges as
necessary[1].  For example, I always used to run VMSMAIL with EXQUOTA
privilege so that the fact that a user had used up all his/her disk quota
did not prevent their receiving mail telling them so :-)

Whilst talking about privileges, VMS's approach of lots of different
privileges seems far safer than Unix's all or nothing (or Windoze's all
all the time:-)

[1] Of course, the person granting those privileges had to have the
necessary authorization him/herself.

-- 
Brian {Hamilton Kelly}                                          bhk@dsl.co.uk
    "We have gone from a world of concentrated knowledge and wisdom to one of
    distributed ignorance.  And we know and understand less while being incr-
    easingly capable."                          Prof. Peter Cochrane, BT Labs

######

From: bhk@dsl.co.uk (Brian {Hamilton Kelly})
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: Mon, 10 Jul 2000 20:11:33 GMT
Organization: Dragonhill Systems Ltd
Message-ID: <963259893snz@dsl.co.uk>
References: <873dlsdrci.fsf@cartman.azz.net> <025a9e75.2d4e185c@usw-ex0108-063.remarq.com> <962890329.979826@shelley.paradise.net.nz>
X-Trace: mail2news.demon.co.uk 963262287 mail2news:12597 mail2news mail2news.demon.co.uk
X-Complaints-To: abuse@demon.net
X-Mail2News-Path: news.demon.net!dsl.demon.co.uk
X-Newsreader: Demon Internet Simple News v1.31
Lines: 27
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news.maxwell.syr.edu!newsfeed.icl.net!dispose.news.demon.net!demon!news.demon.co.uk!demon!mail2news.demon.co.uk!not-for-mail
Xref: chonsp.franklin.ch alt.folklore.computers:59620

In article <962890329.979826@shelley.paradise.net.nz>
           don@news.daedalus.co.nz "Don Stokes" writes:

> martin2305  <m.taylorNOm.SPAM@elsevier.co.uk.invalid> wrote:
> >The ability for an executable image to possess file access
> >privileges *in its own right* (I don't think UNIX OSs have this -
> > please correct me if I'm mistaken). 
> 
> Unix sortof has this, via the setuid & setgid bits.  It's perhaps
> not as general as you might like, but handles most cases moderately
> well.
> 
> I seem to recall the ability to grant executable images access rights
> identifiers as being a feature of VMS that was always going to be
> available in the next release... did they ever get it to work?

Umm, VMS had this from v2.5 onwards (for INSTALLed executables).

(If you're really talking about ACIs/ACLs, these were grantable to images
from v4.2 onwards, IIRC)

-- 
Brian {Hamilton Kelly}                                          bhk@dsl.co.uk
    "We have gone from a world of concentrated knowledge and wisdom to one of
    distributed ignorance.  And we know and understand less while being incr-
    easingly capable."                          Prof. Peter Cochrane, BT Labs

######

From: bhk@dsl.co.uk (Brian {Hamilton Kelly})
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: Mon, 10 Jul 2000 20:14:35 GMT
Organization: Dragonhill Systems Ltd
Message-ID: <963260075snz@dsl.co.uk>
References: <873dlsdrci.fsf@cartman.azz.net> <962890329.979826@shelley.paradise.net.nz> <1ea9msoah4i5ehjg77bqjrv6i0corf4722@4ax.com> <FxArnK.401@spcuna.spc.edu> <962926887.864144@shelley.paradise.net.nz> <FxB09n.232@spcuna.spc.edu>
X-Trace: mail2news.demon.co.uk 963262292 mail2news:12601 mail2news mail2news.demon.co.uk
X-Complaints-To: abuse@demon.net
X-Mail2News-Path: news.demon.net!dsl.demon.co.uk
X-Newsreader: Demon Internet Simple News v1.31
Lines: 24
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!enews.sgi.com!EU.net!blackbush.xlink.net!news0.de.colt.net!colt.net!dispose.news.demon.net!demon!news.demon.co.uk!demon!mail2news.demon.co.uk!not-for-mail
Xref: chonsp.franklin.ch alt.folklore.computers:59616

In article <FxB09n.232@spcuna.spc.edu> terry@gate.tmk.com "Terry Kennedy" writes:

>   * Even the name of the DEC VMS TCP/IP package - "Ultrix Connection" or
> UCX for short - shows how much DEC missed the boat on the Internet - they
> thought people wanted TCP/IP to talk to their Ultrix boxes, not to use the
> Internet. Early UCX was missing NFS, the R-services, and so forth.

Moreover, "The Ultrix Connection" was the most poorly documented, and had
the worst UI, of anything ever written for VMS.  This was down to its
being the only VMS product written by DEC's Ultrix team...

Everyone I knew that wanted TCP/IP either went and bought TGV's MultiNet
(if they could afford it) or used the CMU package (which was somewhat
"buggy").

(Yes, I know that there was Wollongong and other packages also
available.)

-- 
Brian {Hamilton Kelly}                                          bhk@dsl.co.uk
    "We have gone from a world of concentrated knowledge and wisdom to one of
    distributed ignorance.  And we know and understand less while being incr-
    easingly capable."                          Prof. Peter Cochrane, BT Labs

######

From: bhk@dsl.co.uk (Brian {Hamilton Kelly})
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: Mon, 10 Jul 2000 20:19:11 GMT
Organization: Dragonhill Systems Ltd
Message-ID: <963260351snz@dsl.co.uk>
References: <873dlsdrci.fsf@cartman.azz.net> <8_X75.24474$R7.1135952@news-east.usenetserver.com> <3960A21F.916C66B7@tnglwood.demon.co.uk> <3960FD28.889A6941@netcologne.de> <8k5fvj$8t8$1@news.iastate.edu>
X-Trace: mail2news.demon.co.uk 963262298 mail2news:12608 mail2news mail2news.demon.co.uk
X-Complaints-To: abuse@demon.net
X-Mail2News-Path: news.demon.net!dsl.demon.co.uk
X-Newsreader: Demon Internet Simple News v1.31
Lines: 17
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!feed2.news.luth.se!luth.se!newsfeeds.belnet.be!news.belnet.be!colt.net!dispose.news.demon.net!demon!news.demon.co.uk!demon!mail2news.demon.co.uk!not-for-mail
Xref: chonsp.franklin.ch alt.folklore.computers:59615

In article <8k5fvj$8t8$1@news.iastate.edu>
           rhawkins@iastate.edu "Rick Hawkins" writes:

> cryptic?  rm, vi, dd . . . . perfect sense.  Never trust a command
> with more than two letters . . . :)

What, not even tex?

(BTW, why do Unix implementations of TeX not have that for the
executable's name?)

-- 
Brian {Hamilton Kelly}                                          bhk@dsl.co.uk
    "We have gone from a world of concentrated knowledge and wisdom to one of
    distributed ignorance.  And we know and understand less while being incr-
    easingly capable."                          Prof. Peter Cochrane, BT Labs

######

From: bhk@dsl.co.uk (Brian {Hamilton Kelly})
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: Mon, 10 Jul 2000 20:22:19 GMT
Organization: Dragonhill Systems Ltd
Message-ID: <963260539snz@dsl.co.uk>
References: <963206341.272445@server16.cable.com>
X-Trace: mail2news.demon.co.uk 963262304 mail2news:12612 mail2news mail2news.demon.co.uk
X-Complaints-To: abuse@demon.net
X-Mail2News-Path: news.demon.net!dsl.demon.co.uk
X-Newsreader: Demon Internet Simple News v1.31
Lines: 15
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!News.Amsterdam.UnisourceCS!newshunter!cosy.sbg.ac.at!newsfeed.Austria.EU.net!news.nikoma.de!tiscalinetde!news0.de.colt.net!colt.net!dispose.news.demon.net!demon!news.demon.co.uk!demon!mail2news.demon.co.uk!not-for-mail
Xref: chonsp.franklin.ch alt.folklore.computers:59618

In article <963206341.272445@server16.cable.com>
           plyons@ca-unlimited.com "P Linnane" writes:

> Off-topic in any creative sense, but: how the **** do PC programs fill up
> 200MB on a disk ?
> No, seriously, does anyone know ?

Easter Eggs from under-stretched Windoze programmers?

-- 
Brian {Hamilton Kelly}                                          bhk@dsl.co.uk
    "We have gone from a world of concentrated knowledge and wisdom to one of
    distributed ignorance.  And we know and understand less while being incr-
    easingly capable."                          Prof. Peter Cochrane, BT Labs

######

From: Eric Chomko <chomko@IDT.NET>
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: 10 Jul 2000 20:24:35 GMT
Organization: IDT Internet Services
Lines: 16
Message-ID: <8kdbe3$2c9@nnrp4.farm.idt.net>
References: <873dlsdrci.fsf@cartman.azz.net> <877lb4f18g.fsf@litterbox.meowing.net> <cNT75.4138$7%3.326535@news.flash.net> <8k583b$mgh@teal.sni.net> <hnsngr-ya023180000807001837020001@news.pacbell.net> <m3g0pk2b7a.fsf@localhost.localdomain>
NNTP-Posting-Host: u3.farm.idt.net
X-Newsreader: TIN [UNIX 1.3 unoff BETA release 961025]
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!enews.sgi.com!news.idt.net!nntp.farm.idt.net!u3.farm.idt.net!not-for-mail
Xref: chonsp.franklin.ch alt.folklore.computers:59529

Dowe Keller <dowe@localhost.localdomain> wrote:
: hnsngr@sirius.com (Ron Hunsinger) writes:

: > In article <8k583b$mgh@teal.sni.net>, bediger@csn.net (Bruce Ediger) wrote:

: > EBCDIC also, but I didn't mind that. Every once in a while I'd hear rumors
: > of this other character set called ASCII, but it seemed mighty strange to
: > me, with not even any gaps in the codes for the letters. How was I supposed
: > to map character codes to punched cards? Besides, didn't God use EBCDIC?
:                                                     ^^^^^^^^^^^^^^^^^^^^^^
: Actually, that would explain an awful lot of the seeming randomness in the
: universe :-)

On the same token God made the platypus.

Eric

######

From: cbh@REMOVE_THIS.teabag.demon.co.uk (Chris Hedley)
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: 10 Jul 2000 22:10:41 GMT
Organization: teabag
Message-ID: <8kdhl1$15uj$2@teabag.demon.co.uk>
References: <873dlsdrci.fsf@cartman.azz.net>
    <025a9e75.2d4e185c@usw-ex0108-063.remarq.com>
    <396477A1.A772FA58@tnglwood.demon.co.uk> <963259650snz@dsl.co.uk>
NNTP-Posting-Host: teabag.demon.co.uk
X-NNTP-Posting-Host: teabag.demon.co.uk:193.237.4.110
X-Trace: news.demon.co.uk 963267131 nnrp-10:24211 NO-IDENT teabag.demon.co.uk:193.237.4.110
X-Complaints-To: abuse@demon.net
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
NNTP-Posting-Date: 10 Jul 2000 22:10:41 GMT
X-Newsreader: knews 1.0b.0
Lines: 13
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!News.Amsterdam.UnisourceCS!skynet.be!dispose.news.demon.net!demon!news.demon.co.uk!demon!teabag.demon.co.uk!not-for-mail
Xref: chonsp.franklin.ch alt.folklore.computers:59553

In article <963259650snz@dsl.co.uk>,
	bhk@dsl.co.uk (Brian {Hamilton Kelly}) writes:
> Whilst talking about privileges, VMS's approach of lots of different
> privileges seems far safer than Unix's all or nothing (or Windoze's all

This is something of a misconception of Unix' security propagated by
lazy programmers running everything as root[1] when the appropriate
use of groups would be more flexible and secure.

[1] Admittedly, in too many cases there isn't much choice... but that
    still doesn't excuse not looking into the alternatives first IMHO.

Chris.

######

From: jmaynard@thebrain.conmicro.cx (Jay Maynard)
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: 10 Jul 2000 22:21:52 GMT
Organization: Neosoft (using Airnews.net!)
Lines: 13
Message-ID: <EA5A4EA8FAE726A3.44888574B89A9693.9196726B4302A5CF@lp.airnews.net>
X-Orig-Message-ID: <slrn8mkj56.hl5.jmaynard@thebrain.conmicro.cx>
References: <873dlsdrci.fsf@cartman.azz.net> <8_X75.24474$R7.1135952@news-east.usenetserver.com> <3960A21F.916C66B7@tnglwood.demon.co.uk> <3960FD28.889A6941@netcologne.de> <8k5fvj$8t8$1@news.iastate.edu> <963260351snz@dsl.co.uk>
Reply-To: jmaynard@conmicro.cx
Abuse-Reports-To: abuse at airmail.net to report improper postings
NNTP-Proxy-Relay: 204.181.96.50
NNTP-Posting-Time: Mon Jul 10 17:21:52 2000
NNTP-Posting-Host: !YE<r1k-WLFEtp[ (Encoded at Airnews!)
User-Agent: slrn/0.9.5.4 (UNIX)
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!enews.sgi.com!korova.insync.net!solomon.io.com!nntp.giganews.com!cabal14.airnews.net!news.airnews.net!cabal1.airnews.net!news-f.iadfw.net!jmaynard
Xref: chonsp.franklin.ch alt.folklore.computers:59514

On Mon, 10 Jul 2000 20:19:11 GMT, Brian {Hamilton Kelly} <bhk@dsl.co.uk>
wrote:
>What, not even tex?
>(BTW, why do Unix implementations of TeX not have that for the
>executable's name?)

Because Unix commands are case-sensitive, and uppercasing commands is a
pain.

(Note that I do not defend case sensitivity, which I consider the single
greatest user interface botch ever perpetrated on the computing public,
especially since it could have been fixed with a small number of
instructions at Unix v1.)

######

Sender: eric@ruckus.brouhaha.com
From: Eric Smith <eric-no-spam-for-me@brouhaha.com>
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
References: <873dlsdrci.fsf@cartman.azz.net> <8_X75.24474$R7.1135952@news-east.usenetserver.com> <3960A21F.916C66B7@tnglwood.demon.co.uk> <3960FD28.889A6941@netcologne.de> <8k5fvj$8t8$1@news.iastate.edu> <963260351snz@dsl.co.uk>
X-Disclaimer: Everything I write is false.
Organization: Eric Conspiracy Secret Labs
X-Eric-Conspiracy: There is no conspiracy.
Date: 10 Jul 2000 15:46:21 -0700
Message-ID: <qhya39eh8i.fsf@ruckus.brouhaha.com>
Lines: 7
X-Newsreader: Gnus v5.5/Emacs 20.3
NNTP-Posting-Host: ruckus.brouhaha.com
X-Trace: 10 Jul 2000 15:44:42 -0700, ruckus.brouhaha.com
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!newsfeeds.belnet.be!news.belnet.be!newsfeed.direct.ca!look.ca!news.he.net!rn.area.com!news.spies.com!ruckus.brouhaha.com
Xref: chonsp.franklin.ch alt.folklore.computers:59583

bhk@dsl.co.uk (Brian {Hamilton Kelly}) writes:
> (BTW, why do Unix implementations of TeX not have that for the
> executable's name?)

Because Unix hackers don't like to use the shift key any more than
absolutely necessary.  If it was installed as 'TeX', everyone would link
it to 'tex'.

######

Newsgroups: alt.folklore.computers
From: Terry Kennedy <terry@gate.tmk.com>
Subject: Re: Features to save from old OSs
X-Complaints-To: Email abuse@spc.edu if this posting is inappropriate
User-Agent: tin/1.4.2-20000205 ("Possession") (UNIX) (BSD/OS/4.1 (i386))
NNTP-Posting-Host: gate.tmk.com
Organization: St. Peter's College, US
Message-ID: <FxIEKw.HvL@spcuna.spc.edu>
References: <873dlsdrci.fsf@cartman.azz.net> <962890329.979826@shelley.paradise.net.nz> <1ea9msoah4i5ehjg77bqjrv6i0corf4722@4ax.com> <FxArnK.401@spcuna.spc.edu> <962926887.864144@shelley.paradise.net.nz> <FxB09n.232@spcuna.spc.edu> <963260075snz@dsl.co.uk>
X-Trace: spcuna.spc.edu 963278671 22083 terry [204.141.35.61]
Date: Tue, 11 Jul 2000 01:24:32 GMT
Lines: 47
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news.maxwell.syr.edu!howland.erols.net!panix!newsmaster.cc.columbia.edu!ord-feed.news.verio.net!news.verio.net!news.spc.edu!not-for-mail
Xref: chonsp.franklin.ch alt.folklore.computers:59517

Brian {Hamilton Kelly} <bhk@dsl.co.uk> writes:
> Moreover, "The Ultrix Connection" was the most poorly documented, and had
> the worst UI, of anything ever written for VMS.  This was down to its
> being the only VMS product written by DEC's Ultrix team...

  As opposed to DIAGNOSE and Compaq Analyze, which nobody will admit to
writing (recent VMS hardware can't analyze hardware errors with the built-
in tool - ERRFMT - you have to install one of these "cross-platform" night-
mares instead.

  There's a reson UCX is pronounced "Yucks" 8-)

  Actually, the developers were pretty good - it's just that for product
releases, when the 2-minute-egg timer went "ding", management said "Ok,
ship what you have" instead of waiting for the product to be ready. For
support, the customer support center did an incredible job of filtering
problem reports. I managed to see the gibberish that was the result of
them translating my problem report. No wonder the developers couldn't make
sense of the bug reports they were getting.

  At Providence '99, I spoke to the lead developer and was completely con-
vinced that if they let him do the product he wanted to make, it would be
wonderful. However, management was apparently insisting that they fix old
bugs in the soon-to-be-dead code base, when it would have been faster to
finish and release the new version and then answer all the bug reports with
"fixed in the new version".

> Everyone I knew that wanted TCP/IP either went and bought TGV's MultiNet
> (if they could afford it) or used the CMU package (which was somewhat
> "buggy").

  That was later on, when TGV developed an actual pricing strategy. I know
of people who traded things like cases of beer for MultiNet licenses in the
early days.

> (Yes, I know that there was Wollongong and other packages also
> available.)

  Let's see: UCX (renamed TCP/IP Services for VMS; active), MultiNet (sold
to Cisco, sold to Process; active), Process TCPware (active), Wollongong
(renamed Pathway or some such; defunct), CMU/TEK (later just CMU; defunct),
Micom/Interlan (defunct?)

  Any others?

	Terry Kennedy             http://www.tmk.com
        terry@tmk.com             Jersey City, NJ USA

######

Newsgroups: alt.folklore.computers
From: Terry Kennedy <terry@gate.tmk.com>
Subject: Re: Features to save from old OSs
X-Complaints-To: Email abuse@spc.edu if this posting is inappropriate
User-Agent: tin/1.4.2-20000205 ("Possession") (UNIX) (BSD/OS/4.1 (i386))
NNTP-Posting-Host: gate.tmk.com
Organization: St. Peter's College, US
Message-ID: <FxIEtr.I3L@spcuna.spc.edu>
References: <873dlsdrci.fsf@cartman.azz.net> <025a9e75.2d4e185c@usw-ex0108-063.remarq.com> <396477A1.A772FA58@tnglwood.demon.co.uk> <963259650snz@dsl.co.uk>
X-Trace: spcuna.spc.edu 963278990 22083 terry [204.141.35.61]
Date: Tue, 11 Jul 2000 01:29:51 GMT
Lines: 22
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!nyc-news-feed1.bbnplanet.com!news.gtei.net!easynet-tele!easynet.net!news-spur1.maxwell.syr.edu!news.maxwell.syr.edu!howland.erols.net!panix!newsmaster.cc.columbia.edu!ord-feed.news.verio.net!news.verio.net!news.spc.edu!not-for-mail
Xref: chonsp.franklin.ch alt.folklore.computers:59513

Brian {Hamilton Kelly} <bhk@dsl.co.uk> writes:
> VMS executables could readily be given any of the 40+ privileges as
> necessary[1].  For example, I always used to run VMSMAIL with EXQUOTA
> privilege so that the fact that a user had used up all his/her disk quota
> did not prevent their receiving mail telling them so :-)
>
> Whilst talking about privileges, VMS's approach of lots of different
> privileges seems far safer than Unix's all or nothing (or Windoze's all
> all the time:-)

  Of course, you wound up relying on the software authors to not screw
stuff up (or break it in ways that would hurt). For example, current VMS
MAIL images don't properly handle image privs (according to Innosoft - 
see http://www.innosoft.com/app-notes/openvms-mail-v7.html) so installing
MAIL with EXQUOTA now means anyone could SPAWN out of MAIL and have the
EXQUOTA priv.

  Similarly, I remember demonstrating a huge security hole in a bank's
corporate email system - "set editor @tt:" 8-)

	Terry Kennedy             http://www.tmk.com
        terry@tmk.com             Jersey City, NJ USA

######

Newsgroups: alt.folklore.computers
From: Terry Kennedy <terry@gate.tmk.com>
Subject: Re: Features to save from old OSs
X-Complaints-To: Email abuse@spc.edu if this posting is inappropriate
User-Agent: tin/1.4.2-20000205 ("Possession") (UNIX) (BSD/OS/4.1 (i386))
NNTP-Posting-Host: gate.tmk.com
Organization: St. Peter's College, US
Message-ID: <FxJ8CA.ns@spcuna.spc.edu>
References: <873dlsdrci.fsf@cartman.azz.net> <025a9e75.2d4e185c@usw-ex0108-063.remarq.com> <396477A1.A772FA58@tnglwood.demon.co.uk> <963259650snz@dsl.co.uk> <FxIEtr.I3L@spcuna.spc.edu> <slrn8mm1p5.11m.TheCentralScrutinizer.199@C298344-A.arvada1.co.home.com>
X-Trace: spcuna.spc.edu 963317241 264 terry [204.141.35.61]
Date: Tue, 11 Jul 2000 12:07:22 GMT
Lines: 51
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!newsfeeds.belnet.be!news.belnet.be!news-spur1.maxwell.syr.edu!news.maxwell.syr.edu!newsfeed.nyu.edu!newsmaster.cc.columbia.edu!ord-feed.news.verio.net!news.verio.net!news.spc.edu!not-for-mail
Xref: chonsp.franklin.ch alt.folklore.computers:59503

TheCentralScrutinizer.199@pobox.com writes:
> Spawing w/ VMS via a priveledged image doesn't give any of the privs
> to the spawned process.

  Observe:

[priv'd account]

7_SERVER::$ install repl sys$system:mail/priv=exquota

[non-priv'd account]

$ sh proc/priv
[snip]
Authorized privileges:
 NETMBX       TMPMBX
 
Process privileges:
 TMPMBX               may create temporary mailbox
 NETMBX               may create network device

$ mai

MAIL> spawn
$ sh proc/priv
[snip]
Authorized privileges:
 EXQUOTA      NETMBX       TMPMBX
 
Process privileges:
 NETMBX               may create network device
 TMPMBX               may create temporary mailbox

$ set proc/priv=all
%SYSTEM-W-NOTALLPRIV, not all requested privileges authorized
$ sh proc/priv
[snip]
Authorized privileges:
 EXQUOTA      NETMBX       TMPMBX
 
Process privileges:
 EXQUOTA              may exceed disk quota
 NETMBX               may create network device
 TMPMBX               may create temporary mailbox

  This is on V7.2-1, the latest release. Images that aren't written to
manage privs (by disabling them) before spawning will pass those privs
on to spawned subprocesses.

	Terry Kennedy             http://www.tmk.com
        terry@tmk.com             Jersey City, NJ USA

######

From: andrew@cucumber.demon.co.uk (Andrew Gabriel)
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: 11 Jul 2000 20:54:20 GMT
Organization: home
Message-ID: <8kg1hs$ms@cucumber.demon.co.uk>
References: <873dlsdrci.fsf@cartman.azz.net> <8k71cn$2a9@cucumber.demon.co.uk> <8kd5gh$45m$1@news.iastate.edu>
NNTP-Posting-Host: cucumber
X-NNTP-Posting-Host: cucumber.demon.co.uk:158.152.58.86
X-Trace: news.demon.co.uk 963352664 nnrp-09:26821 NO-IDENT cucumber.demon.co.uk:158.152.58.86
X-Complaints-To: abuse@demon.net
X-Newsreader: knews 0.9.6
Lines: 23
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!isdnet!newsfeed.icl.net!colt.net!dispose.news.demon.net!demon!news.demon.co.uk!demon!cucumber.demon.co.uk!usenet
Xref: chonsp.franklin.ch alt.folklore.computers:59662

In article <8kd5gh$45m$1@news.iastate.edu>,
	rhawkins@iastate.edu (Rick Hawkins) writes:
>In article <8k71cn$2a9@cucumber.demon.co.uk>,
>Andrew Gabriel <andrew@cucumber.demon.co.uk> wrote:
>>In article <873dlsdrci.fsf@cartman.azz.net>,
>>       Adam Sampson <azz@gnu.org> writes:
>>
>>Kernel modules and device drivers to operate in separate protected
>>environments which:
>>a) Don't lose whole system if they crash (at least, configurable per module)
>
>So a little flag to set, $CRASH_REST_OF_SYSTEM_WHILE_AT_IT

Well, it was pretty much like that. There were some subtle extra
options relating to whom module crashes were reported if not system
fatal, and you could choose different handling for errors occuring
whilst modules were still starting up, verses errors once system
into normal running state.

-- 
Andrew Gabriel
Consultant Software Engineer

######

From: rhawkins@iastate.edu (Rick Hawkins)
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: 11 Jul 2000 22:09:27 GMT
Organization: Iowa State University, Ames, Iowa USA
Lines: 21
Message-ID: <8kg5un$g7q$1@news.iastate.edu>
References: <873dlsdrci.fsf@cartman.azz.net> <3960FD28.889A6941@netcologne.de> <8k5fvj$8t8$1@news.iastate.edu> <963260351snz@dsl.co.uk>
NNTP-Posting-Host: pv2079.vincent.iastate.edu
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!News.Amsterdam.UnisourceCS!skynet.be!logbridge.uoregon.edu!newsrelay.iastate.edu!news.iastate.edu!rhawkins
Xref: chonsp.franklin.ch alt.folklore.computers:59646

In article <963260351snz@dsl.co.uk>,
Brian {Hamilton Kelly} <bhk@dsl.co.uk> wrote:
>In article <8k5fvj$8t8$1@news.iastate.edu>
>           rhawkins@iastate.edu "Rick Hawkins" writes:

>> cryptic?  rm, vi, dd . . . . perfect sense.  Never trust a command
>> with more than two letters . . . :)

>What, not even tex?

that *does* make it suspect :)  It gets partway off the hook for
not being a system command :)

hawk


-- 
Richard E. Hawkins
dochawk@psu.edu [regardless of where the message says it comes from]

These opinions will not be those of  Penn State until they pay my retainer.

######

From: rhawkins@iastate.edu (Rick Hawkins)
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: 11 Jul 2000 22:11:25 GMT
Organization: Iowa State University, Ames, Iowa USA
Lines: 21
Message-ID: <8kg62d$g98$1@news.iastate.edu>
References: <873dlsdrci.fsf@cartman.azz.net> <8k71of$qdd$4@bob.news.rcn.net> <8kd5ab$4vr$1@news.iastate.edu> <8kf0ih$9sm$1@bob.news.rcn.net>
NNTP-Posting-Host: pv2079.vincent.iastate.edu
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!news.datacomm.ch!newsmaster-01.atnet.at!atnet.at!newsrouter.chello.at!newsfeed01.sul.t-online.de!newsfeed00.sul.t-online.de!t-online.de!news-spur1.maxwell.syr.edu!news.maxwell.syr.edu!newsfeed.stanford.edu!headwall.stanford.edu!unlnews.unl.edu!newsrelay.iastate.edu!news.iastate.edu!rhawkins
Xref: chonsp.franklin.ch alt.folklore.computers:59641

In article <8kf0ih$9sm$1@bob.news.rcn.net>,  <jmfbahciv@aol.com> wrote:


>>hawk, probably misreading this all, and deliberately avoiding the 
>>obvious windows slam

>I'd never slam the windows on you, Hawk.  I'd be afraid that
>I'd miss a really good line of humor :-).  Glad to see you
>back on-line...hope it lasts.

It should stick around this time :)  At the moment I've got
debian installing on this little machine, because FreeBSD just
didn't fit very well.  *sigh*

hawk

-- 
Richard E. Hawkins
dochawk@psu.edu [regardless of where the message says it comes from]

These opinions will not be those of  Penn State until they pay my retainer.

######

Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
References: <963206341.272445@server16.cable.com> <963260539snz@dsl.co.uk>
From: Dowe Keller <dowe@localhost.localdomain>
Date: 11 Jul 2000 21:04:24 -0700
Message-ID: <m34s5wm1tj.fsf@localhost.localdomain>
Lines: 19
X-Newsreader: Gnus v5.5/Emacs 20.3
NNTP-Posting-Host: 209.234.196.87
X-Original-NNTP-Posting-Host: 209.234.196.87
X-Trace: 11 Jul 2000 21:07:48 -0700, 209.234.196.87
Organization: news.sierratel.com
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!enews.sgi.com!news-peer.gip.net!news.gsl.net!gip.net!europa.netcrusader.net!208.184.7.66!newsfeed.skycache.com!Cidera!209.155.26.10!news.sierratel.com!209.234.196.87
Xref: chonsp.franklin.ch alt.folklore.computers:59682

bhk@dsl.co.uk (Brian {Hamilton Kelly}) writes:

> In article <963206341.272445@server16.cable.com>
>            plyons@ca-unlimited.com "P Linnane" writes:
> 
> > Off-topic in any creative sense, but: how the **** do PC programs fill up
> > 200MB on a disk ?
> > No, seriously, does anyone know ?
> 
> Easter Eggs from under-stretched Windoze programmers?
                   ^^^^^^^^^^^^^^^

What does Mocrosoft have a rack? /:=)

-- 
dowe@sierratel.com
This is the theory that Jack built.
This is the flaw that lay in the theory that Jack built.
This is the palpable verbal haze that hid the flaw that lay in...

######

Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
References: <873dlsdrci.fsf@cartman.azz.net> <3960FD28.889A6941@netcologne.de> <8k5fvj$8t8$1@news.iastate.edu> <963260351snz@dsl.co.uk>
Organization: None
X-Newsreader: trn 4.0-test72 (19 April 1999)
From: kragen@dnaco.net (Kragen Sitaker)
Lines: 15
Message-ID: <V9Sa5.2383$sR.10618@east3.usenetserver.com>
X-Abuse-Info: Please be sure to forward a copy of ALL headers
X-Abuse-Info: Otherwise we will be unable to process your complaint properly
X-Complaints-To: support@usenetserver.com
NNTP-Posting-Date: Wed, 12 Jul 2000 00:07:17 EDT
Date: Wed, 12 Jul 2000 04:07:17 GMT
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!HSNX.atgi.net!cyclone2.usenetserver.com!news-out.usenetserver.com!cyclone1.usenetserver.com!cyclone1.usenetserver.com!east3.usenetserver.com.POSTED!not-for-mail
Xref: chonsp.franklin.ch alt.folklore.computers:59677

In article <963260351snz@dsl.co.uk>,
Brian {Hamilton Kelly} <bhk@dsl.co.uk> wrote:
>What, not even tex?
>
>(BTW, why do Unix implementations of TeX not have that for the
>executable's name?)

Mixed-case names in a case-sensitive namespace (such as Unix, C, C++,
or Java) are inexcusable.  Monocase names make case-sensitivity
bearable.
-- 
<kragen@pobox.com>       Kragen Sitaker     <http://www.pobox.com/~kragen/>
Perilous to all of us are the devices of an art deeper than we ourselves        
possess.                                                                        
                -- Gandalf the Grey [J.R.R. Tolkien, "Lord of the Rings"]       
######

From: jmfbahciv@aol.com
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: Sat, 15 Jul 00 09:35:44 GMT
Organization: UltraNet Communications, Inc.
Lines: 45
Message-ID: <8kplk8$4h2$1@bob.news.rcn.net>
References: <873dlsdrci.fsf@cartman.azz.net> <962890329.979826@shelley.paradise.net.nz> <1ea9msoah4i5ehjg77bqjrv6i0corf4722@4ax.com> <FxArnK.401@spcuna.spc.edu> <962926887.864144@shelley.paradise.net.nz> <FxB09n.232@spcuna.spc.edu> <963260075snz@dsl.co.uk> <FxIEKw.HvL@spcuna.spc.edu>
X-Trace: LQFvLAz2Mk0vvscdBMk50EzrWeuCNSFlAxl8V2xxXZs=
X-Complaints-To: abuse@rcn.com
NNTP-Posting-Date: 15 Jul 2000 12:32:08 GMT
X-Newsreader:  News Xpress Version 1.0 Beta #4
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news.maxwell.syr.edu!feed1.news.rcn.net!rcn!207-172-102-250
Xref: chonsp.franklin.ch alt.folklore.computers:59807

In article <FxIEKw.HvL@spcuna.spc.edu>,
   Terry Kennedy <terry@gate.tmk.com> wrote:
>Brian {Hamilton Kelly} <bhk@dsl.co.uk> writes:
>> Moreover, "The Ultrix Connection" was the most 
>> poorly documented, and had
>> the worst UI, of anything ever written for VMS.  This was down to its
>> being the only VMS product written by DEC's Ultrix team...
>
>  As opposed to DIAGNOSE and Compaq Analyze, which nobody will admit to
>writing (recent VMS hardware can't analyze hardware errors with the built-
>in tool - ERRFMT - you have to install one of these 
>"cross-platform" night-
>mares instead.
>
>  There's a reson UCX is pronounced "Yucks" 8-)
>
>  Actually, the developers were pretty good - it's just that for product
>releases, when the 2-minute-egg timer went "ding", management said "Ok,
>ship what you have" instead of waiting for the product to be ready.

This was at DEC?  If that was the case, then they threw out
the criteria that had been established in the Project Notebook.
We (PDP-10 land) were never allowed to do that.

> For
>support, the customer support center did an incredible job of filtering
>problem reports. I managed to see the gibberish that was the result of
>them translating my problem report. No wonder the developers couldn't make
>sense of the bug reports they were getting.
>
>  At Providence '99, I spoke to the lead developer and was completely con-
>vinced that if they let him do the product he wanted to make, it would be
>wonderful. However, management was apparently insisting that they fix old
>bugs in the soon-to-be-dead code base, when it would have been faster to
>finish and release the new version and then answer all the bug 
>reports with "fixed in the new version".

That's only useful if the new version can be installed on the 
hardware the old version supported.

<snip>

/BAH

Subtract a hundred and four for e-mail.

######

From: jmfbahciv@aol.com
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: Sun, 16 Jul 00 08:24:59 GMT
Organization: UltraNet Communications, Inc.
Lines: 52
Message-ID: <8ks5rt$i5c$2@bob.news.rcn.net>
References: <873dlsdrci.fsf@cartman.azz.net> <962890329.979826@shelley.paradise.net.nz> <1ea9msoah4i5ehjg77bqjrv6i0corf4722@4ax.com> <FxArnK.401@spcuna.spc.edu> <962926887.864144@shelley.paradise.net.nz> <FxB09n.232@spcuna.spc.edu> <963260075snz@dsl.co.uk> <FxIEKw.HvL@spcuna.spc.edu> <8kplk8$4h2$1@bob.news.rcn.net> <qhhf9qvoh5.fsf@ruckus.brouhaha.com>
X-Trace: 3yX6Trnkb9N9rs5jkl4W3MhpYYc7BLbxDutlrh5JkFQ=
X-Complaints-To: abuse@rcn.com
NNTP-Posting-Date: 16 Jul 2000 11:21:33 GMT
X-Newsreader:  News Xpress Version 1.0 Beta #4
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news.maxwell.syr.edu!feed1.news.rcn.net!rcn!207-172-102-126
Xref: chonsp.franklin.ch alt.folklore.computers:59872

In article <qhhf9qvoh5.fsf@ruckus.brouhaha.com>,
   Eric Smith <eric-no-spam-for-me@brouhaha.com> wrote:
>Terry Kennedy <terry@gate.tmk.com> wrote:
>> Actually, the developers were pretty good - it's just that for product
>> releases, when the 2-minute-egg timer went "ding", management said "Ok,
>> ship what you have" instead of waiting for the product to be ready.
>
>jmfbahciv@aol.com wrote:
>> This was at DEC?  If that was the case, then they threw out
>> the criteria that had been established in the Project Notebook.
>> We (PDP-10 land) were never allowed to do that.
>                              ^^^^^^^
>I think you mean              forced
>
>
>I've never heard of good engineers being "allowed" to ship software that
>isn't done, and actually choosing to do so.  But I've heard *many* cases
>of engineers being forced to ship before they're ready, and usually that
>results in disaster for everyone concerned.

That may have happened in TOPS-10's younger years (early 70s
and before) since I wasn't a part of that business then.  But
I remember we would be delayed by months because of "objections"
from members of groups that had been created because of the
Project Notebook strictures.  Getting an OS product out of the
door at DEC was so damn painful that I'm not sure how we stood
now that I'm looking back on it.


>
>A friend tells a tale of chip designers being forced to tape out before
>they were ready, and the chip took two extra spins and was six months
>late as a result.  The manager apparently actually learned from this
>experience and now lets the engineers decide when it's time to tape out.
>
>Software is a bit more malleable than silicon, but it still can be
>difficult and expensive to correct the damage from a premature release.

Well, there were also cases where the salemen sold a piece of 
newly developed hardware.  Another problem that JMF and TW had
was getting such a piece hooked up to our machine so they could
write, debug and test both the code and hardware.  One of the
things that we could never pound into the hardware types is that
the hardware is just an expensive boat anchor unless there's
some software that allows the customer to use it.

There were many times when the first hardware ship went to
a customer rather than the developer.

/BAH

Subtract a hundred and four for e-mail.

######

Sender: eric@ruckus.brouhaha.com
From: Eric Smith <eric-no-spam-for-me@brouhaha.com>
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
References: <873dlsdrci.fsf@cartman.azz.net> <962890329.979826@shelley.paradise.net.nz> <1ea9msoah4i5ehjg77bqjrv6i0corf4722@4ax.com> <FxArnK.401@spcuna.spc.edu> <962926887.864144@shelley.paradise.net.nz> <FxB09n.232@spcuna.spc.edu> <963260075snz@dsl.co.uk> <FxIEKw.HvL@spcuna.spc.edu> <8kplk8$4h2$1@bob.news.rcn.net>
X-Disclaimer: Everything I write is false.
Organization: Eric Conspiracy Secret Labs
X-Eric-Conspiracy: There is no conspiracy.
Date: 15 Jul 2000 18:43:50 -0700
Message-ID: <qhhf9qvoh5.fsf@ruckus.brouhaha.com>
Lines: 25
X-Newsreader: Gnus v5.5/Emacs 20.3
NNTP-Posting-Host: ruckus.brouhaha.com
X-Trace: 15 Jul 2000 18:43:08 -0700, ruckus.brouhaha.com
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news-fra1.dfn.de!unlisys!news.snafu.de!news-spur1.maxwell.syr.edu!news.maxwell.syr.edu!newsfeed.stanford.edu!paloalto-snf1.gtei.net!news.gtei.net!webtv.net!news.spies.com!ruckus.brouhaha.com
Xref: chonsp.franklin.ch alt.folklore.computers:59903

Terry Kennedy <terry@gate.tmk.com> wrote:
> Actually, the developers were pretty good - it's just that for product
> releases, when the 2-minute-egg timer went "ding", management said "Ok,
> ship what you have" instead of waiting for the product to be ready.

jmfbahciv@aol.com wrote:
> This was at DEC?  If that was the case, then they threw out
> the criteria that had been established in the Project Notebook.
> We (PDP-10 land) were never allowed to do that.
                              ^^^^^^^
I think you mean              forced


I've never heard of good engineers being "allowed" to ship software that
isn't done, and actually choosing to do so.  But I've heard *many* cases
of engineers being forced to ship before they're ready, and usually that
results in disaster for everyone concerned.

A friend tells a tale of chip designers being forced to tape out before
they were ready, and the chip took two extra spins and was six months
late as a result.  The manager apparently actually learned from this
experience and now lets the engineers decide when it's time to tape out.

Software is a bit more malleable than silicon, but it still can be
difficult and expensive to correct the damage from a premature release.

######

From: jcmorris@jmorris-pc.MITRE.ORG (Joe Morris)
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: 17 Jul 2000 13:25:54 GMT
Organization: The MITRE Corporation
Lines: 14
Message-ID: <8kv1h2$p1q$1@top.mitre.org>
References: <873dlsdrci.fsf@cartman.azz.net> <962890329.979826@shelley.paradise.net.nz> <1ea9msoah4i5ehjg77bqjrv6i0corf4722@4ax.com> <FxArnK.401@spcuna.spc.edu> <962926887.864144@shelley.paradise.net.nz> <FxB09n.232@spcuna.spc.edu> <963260075snz@dsl.co.uk> <FxIEKw.HvL@spcuna.spc.edu> <8kplk8$4h2$1@bob.news.rcn.net> <qhhf9qvoh5.fsf@ruckus.brouhaha.com> <8ks5rt$i5c$2@bob.news.rcn.net>
NNTP-Posting-Host: jmorris-pc.mitre.org
X-Trace: top.mitre.org 963840354 25658 128.29.251.13 (17 Jul 2000 13:25:54 GMT)
X-Complaints-To: usenet@news.mitre.org
NNTP-Posting-Date: 17 Jul 2000 13:25:54 GMT
X-Newsreader: NN version 6.5.0 (NOV)
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!newsfeeds.belnet.be!news.belnet.be!news-spur1.maxwell.syr.edu!news.maxwell.syr.edu!arclight.uoregon.edu!news.tufts.edu!blanket.mitre.org!news.mitre.org!jmorris-pc.MITRE.ORG!jcmorris
Xref: chonsp.franklin.ch alt.folklore.computers:59858

jmfbahciv@aol.com writes:

>                           Another problem that JMF and TW had
>was getting such a piece hooked up to our machine so they could
>write, debug and test both the code and hardware.  

One of Grace Hopper's stories from her early days (she described
herself as "the third programmer") was working on programming
the computer during the day, after which the engineers would come in
at night and change the hardware design.  A good example of an
iterative process, or perhaps a better example of the chicken and
the egg fighting to see which one should come first.

Joe Morris

######

From: "Charlie Gibbs" <cgibbs@sky.bus.com>
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: 17 Jul 00 10:49:08 -0800
Organization: http://extra.newsguy.com
Lines: 24
Message-ID: <1080.233T408T6493445@sky.bus.com>
References: <FxArnK.401@spcuna.spc.edu> <962926887.864144@shelley.paradise.net.nz> <FxB09n.232@spcuna.spc.edu> <963260075snz@dsl.co.uk> <FxIEKw.HvL@spcuna.spc.edu> <8kplk8$4h2$1@bob.news.rcn.net> <qhhf9qvoh5.fsf@ruckus.brouhaha.com>
NNTP-Posting-Host: p-059.newsdawg.com
X-Newsreader: THOR 2.5a (Amiga;TCP/IP)
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!News.Amsterdam.UnisourceCS!newshunter!cosy.sbg.ac.at!news.cs.utwente.nl!newnews.netizen.com.au!nntp.cs.ubc.ca!news-spur1.maxwell.syr.edu!news.maxwell.syr.edu!newsfeed.gamma.ru!Gamma.RU!pln-e!spln!dex!extra.newsguy.com!newsp.newsguy.com!news2
Xref: chonsp.franklin.ch alt.folklore.computers:59948

In article <qhhf9qvoh5.fsf@ruckus.brouhaha.com>
eric-no-spam-for-me@brouhaha.com (Eric Smith) writes:

>Software is a bit more malleable than silicon, but it still can be
>difficult and expensive to correct the damage from a premature release.

Often it's impossible.  One of the reasons we have so much buggy
software is that other software is exploiting the bug.  Correcting
the bug will break that other software, so it can never be fixed.
When porting or emulating existing software, the bugs have to be
brought across - hence the term "bug-compatible".

I've always been very cautious when designing anything, however
trivial, because I know that I'll probably have to live with it
forever.  Or, to paraphrase the old saying: "If at first you don't
succeed, you might as well forget it."  (Thanks for that one goes to
Robert L. Glass, a.k.a. "Miles Benson", whose Computerworld columns
contained lessons that remain unlearned to this day by too many
people who should know better.)

--
cgibbs@sky.bus.com (Charlie Gibbs)
Remove the first period after the "at" sign to reply.

######

From: jmfbahciv@aol.com
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: Tue, 18 Jul 00 08:00:00 GMT
Organization: UltraNet Communications, Inc.
Lines: 24
Message-ID: <8l1d5j$f0u$2@bob.news.rcn.net>
References: <873dlsdrci.fsf@cartman.azz.net> <962890329.979826@shelley.paradise.net.nz> <1ea9msoah4i5ehjg77bqjrv6i0corf4722@4ax.com> <FxArnK.401@spcuna.spc.edu> <962926887.864144@shelley.paradise.net.nz> <FxB09n.232@spcuna.spc.edu> <963260075snz@dsl.co.uk> <FxIEKw.HvL@spcuna.spc.edu> <8kplk8$4h2$1@bob.news.rcn.net> <qhhf9qvoh5.fsf@ruckus.brouhaha.com> <8ks5rt$i5c$2@bob.news.rcn.net> <8kv1h2$p1q$1@top.mitre.org>
X-Trace: wd/PN3IhX1m+ZGoAPOWxIGzH/OrQe3eInU2/HHQofyQ=
X-Complaints-To: abuse@rcn.com
NNTP-Posting-Date: 18 Jul 2000 10:56:51 GMT
X-Newsreader:  News Xpress Version 1.0 Beta #4
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!News.Amsterdam.UnisourceCS!skynet.be!europa.netcrusader.net!207.172.3.37!feed1.news.rcn.net!rcn!207-172-255-3
Xref: chonsp.franklin.ch alt.folklore.computers:59955

In article <8kv1h2$p1q$1@top.mitre.org>,
   jcmorris@jmorris-pc.MITRE.ORG (Joe Morris) wrote:
>jmfbahciv@aol.com writes:
>
>>                           Another problem that JMF and TW had
>>was getting such a piece hooked up to our machine so they could
>>write, debug and test both the code and hardware.  
>
>One of Grace Hopper's stories from her early days (she described
>herself as "the third programmer") was working on programming
>the computer during the day, after which the engineers would come in
>at night and change the hardware design.  A good example of an
>iterative process, or perhaps a better example of the chicken and
>the egg fighting to see which one should come first.

Oh, yeah :-).  I think that problem was solved by not allowing
the hardware types to touch the software machine and guarding
it with heavy listings.  And then, of course, there's the
introduction of the diagnostics group; an event that happens
once a company figures out how to produce machines.

/BAH

Subtract a hundred and four for e-mail.

######

From: jmfbahciv@aol.com
Newsgroups: alt.folklore.computers
Subject: Re: Features to save from old OSs
Date: Tue, 18 Jul 00 08:01:10 GMT
Organization: UltraNet Communications, Inc.
Lines: 18
Message-ID: <8l1d7o$f0u$3@bob.news.rcn.net>
References: <FxArnK.401@spcuna.spc.edu> <962926887.864144@shelley.paradise.net.nz> <FxB09n.232@spcuna.spc.edu> <963260075snz@dsl.co.uk> <FxIEKw.HvL@spcuna.spc.edu> <8kplk8$4h2$1@bob.news.rcn.net> <qhhf9qvoh5.fsf@ruckus.brouhaha.com> <1080.233T408T6493445@sky.bus.com>
X-Trace: GBXH35g4ypRdd6x8VNz9jAAZFZAIMeWAcgJ0VGP14G0=
X-Complaints-To: abuse@rcn.com
NNTP-Posting-Date: 18 Jul 2000 10:58:00 GMT
X-Newsreader:  News Xpress Version 1.0 Beta #4
Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!newscore.univie.ac.at!194.25.134.126.MISMATCH!newsfeed01.sul.t-online.de!newsfeed00.sul.t-online.de!t-online.de!diablo.theplanet.net!europa.netcrusader.net!207.172.3.37!feed1.news.rcn.net!rcn!207-172-255-3
Xref: chonsp.franklin.ch alt.folklore.computers:59956

In article <1080.233T408T6493445@sky.bus.com>,
   "Charlie Gibbs" <cgibbs@sky.bus.com> wrote:
>In article <qhhf9qvoh5.fsf@ruckus.brouhaha.com>
>eric-no-spam-for-me@brouhaha.com (Eric Smith) writes:
>
>>Software is a bit more malleable than silicon, but it still can be
>>difficult and expensive to correct the damage from a premature release.
>
>Often it's impossible.  One of the reasons we have so much buggy
>software is that other software is exploiting the bug.  

Which is the reason I don't like doing OS developement in HLLs.

<snip>

/BAH

Subtract a hundred and four for e-mail.