Newsgroups: alt.sys.pdp8,alt.sys.pdp10,alt.folklore.computers From: Christopher C Stacy Subject: Filenames, shells, remote file access [was: ooh, a *real* flamewar :) Sender: news@world.std.com (Mr Usenet Himself) Message-ID: Date: Fri, 24 Mar 2000 07:43:23 GMT References: <8bbm3j$8in$1@pravda.ucr.edu> <38da49e6_2@news.wizvax.net> <8beb7i$707$1@citadel.in.taronga.com> NNTP-Posting-Host: world.std.com Organization: The World @ Software Tool & Die X-Newsreader: Gnus v5.7/Emacs 20.5 Lines: 125 Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!isdnet!netnews.com!europa.netcrusader.net!63.208.208.143!feed2.onemain.com!feed1.onemain.com!uunet!ffx.uu.net!world!news Xref: chonsp.franklin.ch alt.folklore.computers:52778 >>>>> On 24 Mar 2000 00:00:50 GMT, Peter da Silva ("Peter") writes: Peter> In article <38da49e6_2@news.wizvax.net>, John Wilson wrote: >> OK how about the flip side, how come no one seems to have been interested in >> writing a command processor replacement for T10/T20/ITS that *did* expand *? Peter> The command processor on TOPS was a monitor, not a shell. It couldn't be Peter> conveniently replaced. The user's command processor on ITS was a user program, and people did write replacements. I think under UNIX someone has recently implemented a Lisp (well, Scheme, actually) shell ... ITS had those of things. Of course, the command processor on the Lisp Machine understood all about file names and wildcards and filename completion and such, not only for the native file system, but also for the all the other operating systems: ITS, several different flavors of Unix with different rules, DOS, TOPS-20, VMS, Multics, Macintosh, and others that I can't remember. File name handling was provided by the operating system. (This operating system was rather different than UNIX or Windows, but rather than thinking "monolithic kernel", you should think more like "dynamic libraries".) Transparent file access (via FTP, NFS, NFILE, or any other protocol -- automatically selected) was also built into the system, so any application or user command could open or manipulate any file on any system simply by referring to the file. The system would figure out what kind of operating system the file server was running, what the best protocols were to access the file, and take care of connecting and running it all under the hood for you. The file name was not mapped to a standard convention - you used the remote file system's syntax. So if you wanted a file "/home/cstacy/foobar" that lived on some Unix box named BART, the file "Open" library call would take this string: BART:/home/cstacy/Foobar.txt For a similarly named file on TOPS-20, it would be LISA:FOOBAR.TXT and on the native LispM file system (which was sort of like Multics, except that it had types and versions and other goodies) you would say HOMER:>CStacy>Foobar.text By the way, LMFS was like what NTFS does with case: it preserves the input case you specified, but the case is not significant. If you were typing to the command processor, you would type the same thing, and it would know how to do filename completion in the appropriate syntax. In addition to basic syntax rules and restrictions, the system also knew about whether the (remote or local) file system supported types, version numbers, links, all the different kinds of file properties for devices and files and directories on each file system, which operations like wildcards had to be emulated and which were supported by the remote file protocol, whether there were file links, etc. etc. etc. The system could also merge pathnames properly, even across different types of files systems; this made it possible for filename defaulting to work. For one example, you could do: :Copy File LISA:FOOBAR.TXT BART:/home/cstacy/ and have it do the right thing (create a file named "foobar.txt"). That example doesn't show it, but I am pretty sure it knew that even though UNIX didn't have file types, that the right thing would to automatically convert for a system that used "foo.text" into the typical "foo.txt" when doing that kind of filename merging. It had rules for handling the ideosyncrocies of each filesystem. Understanding all about wildcards was really handy for copying files around between two other operating systems. It even worked great to do it when the targets were two UNIX filesystems, even though it would have required writing a very complicated shell or perl program to do it on those systems natively! The basis for this was not simply pattern matching, but rather algorithms which actually knew what was going on with the file names and knew about the different operating systems. Also, it did *not* give the user full regexp matching support, only partial, so you could not say "'f' followed by any three characters then two numeric digits and the string 'you'". But given what it *did* do, there seemed no practical need -- nobody ever asked for anything like that. But if it came up, that could be added to the system very easily (in a few days), and then all the applications and user interfaces would have been able to exploit it. There was also a provision for creating a mapping into a standard generic filename syntax, which was useful for certain application delivery issues. This involved telling the system that there was a fake ("logical") host, say, nambed BURNS, and then telling it that BURNS was really LISA, and that the toplevel directory was "/home". So when you said "BURNS:CSTACY;FOOBAR.TEXT" it was turned into "LISA:/home/cstacy/foobar". This was useful if you had a site where the actual location of the files moved around from server to server periodically (such as when the server got too full). For example, you could always refer to the file SYS:NET;TCP.LISP to get the sources to the TCP implementation, and the system would automatically find the right file server (and pattern-match specified directory structure). All of these features I've described in this entire message were part of the system -- free and automatic and invisible to all programs and all user interfaces on the system. It's all part of the service of the normal calls such as OPEN and RENAME and ACCEPT-PATHNAME (completing user input), etc. No "mounting" or anything like that was required on the part of the user. There was a distributed database at the site which contained the configuration information - mostly this said which hosts ran which operating systems, and what protocols were available. The system also supported DNS and NIS and integrated the information from those sources. Everything was extensible, of course. That's one of the main points of the operating system. So If you had a new file system on some operating system, or even a new model for how an existing file system should be viewed or used, you could implement support for it. The system is multiple-inheritence object oriented (and that's very easy and elegant to program in Lisp, unlike in C++), so you could mix and match behaviours from existing file system classes and methods and specialize them as needed. It would take an experienced system programmer somewhere between a couple hours and a couple of days to bring up support for a new kind of file system. This is just a quickie-tip-of-the-iceberg-description of the coolness, of course. It is so far ahead of what the currently popular operating systems do that I am not at all confident that I am getting it across to you in such a terse explanation. (Whatever you're probably thinking based on what I hve told you -- most likely it was in reality a lot better than that and I have not described thoroughly enough.) ###### From: peter@taronga.com (Peter da Silva) Newsgroups: alt.sys.pdp8,alt.sys.pdp10,alt.folklore.computers Subject: Re: Filenames, shells, remote file access [was: ooh, a *real* flamewar :) Date: 24 Mar 2000 15:02:49 GMT Organization: TSS Inc. Lines: 43 Message-ID: <8bg02p$1b2k$1@citadel.in.taronga.com> References: <38da49e6_2@news.wizvax.net> <8beb7i$707$1@citadel.in.taronga.com> NNTP-Posting-Host: citadel.in.taronga.com X-Trace: citadel.in.taronga.com 953910169 44116 10.0.0.43 (24 Mar 2000 15:02:49 GMT) X-Complaints-To: usenet@taronga.com NNTP-Posting-Date: 24 Mar 2000 15:02:49 GMT X-Newsreader: trn 4.0-test72 (19 April 1999) Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!enews.sgi.com!newsfeed.berkeley.edu!newsfeed.stanford.edu!news-proxy.baileynm.com!citadel.in.taronga.com!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:52776 In article , Christopher C Stacy wrote: >The file name was not mapped to a standard convention - you used the remote >file system's syntax. So if you wanted a file "/home/cstacy/foobar" that >lived on some Unix box named BART, the file "Open" library call would take >this string: > > BART:/home/cstacy/Foobar.txt > >For a similarly named file on TOPS-20, it would be > > LISA:FOOBAR.TXT I'm somewhat boggled. It's hard enough dealing with one file naming convention at a time... I've written libraries to do that, to make it easier to write programs that would run on UNIX, AmigaOS, VMS, and DOS. Trying to write a program that did the right thing with a list of files that might have completely different syntaxes seems impossible. How did it deal with this? The only think I can think of that would work would be to know what kind of system each was... and that would break down if you had to deal with file names on machines that weren't currently available on the network. Or did you simply not do that? By the way, there were a couple of early UNIX remote file systems that came much closer to what you're talking about. Rather than step through the remote file system step by step, the "remote" part of a file name was simply passed to the other and for translation. File systems on the network used a standard syntax, though, so when you looked at them from a DOS box they had DOS syntax so DOS programs didn't think UNIX filenames were command options, and UNIX shells saw UNIX paths so they didn't treat VMS directories as regular expressions. If you're trying to make a system that works transparently from all sides, in a heterogenous environment, you can't just optimise for one operating system. -- This is The Reverend Peter da Silva's Boring Sig File - there are no references to Wolves, Kibo, Discordianism, or The Church of the Subgenius in this document Executive Vice President, Corporate Communications, Entropy Gradient Reversals. ###### Newsgroups: alt.sys.pdp8,alt.sys.pdp10,alt.folklore.computers From: Christopher C Stacy Subject: Re: Filenames, shells, remote file access [was: ooh, a *real* flamewar :) Sender: news@world.std.com (Mr Usenet Himself) Message-ID: Date: Fri, 24 Mar 2000 22:43:41 GMT References: <38da49e6_2@news.wizvax.net> <8beb7i$707$1@citadel.in.taronga.com> <8bg02p$1b2k$1@citadel.in.taronga.com> NNTP-Posting-Host: world.std.com Organization: The World @ Software Tool & Die X-Newsreader: Gnus v5.7/Emacs 20.5 Lines: 89 Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!news.datacomm.ch!newscore.gigabell.net!news0.de.colt.net!colt.net!ayres.ftech.net!news.ftech.net!news-hub.cableinet.net!newsfeed.wirehub.nl!tank.news.pipex.net!pipex!uunet!zur.uu.net!ffx.uu.net!world!news Xref: chonsp.franklin.ch alt.folklore.computers:52873 >>>>> On 24 Mar 2000 15:02:49 GMT, Peter da Silva ("Peter") writes: Peter> In article , Peter> Christopher C Stacy wrote: >> The file name was not mapped to a standard convention - you used the remote >> file system's syntax. So if you wanted a file "/home/cstacy/foobar" that >> lived on some Unix box named BART, the file "Open" library call would take >> this string: >> >> BART:/home/cstacy/Foobar.txt >> >> For a similarly named file on TOPS-20, it would be >> >> LISA:FOOBAR.TXT Peter> I'm somewhat boggled. Peter> It's hard enough dealing with one file naming convention at a time.. The point is that the operating system allowed users to keep their files on whatever kind of remote filesystem they preferred, and refer to them in the way that would be most familiar to the user. If they had a bunch of UNIX boxes that they liked, then they could just access those files using the regular UNIX file names. If you had several different kinds of file servers in the shop, and you did some of your development on Windows and some on UNIX and kept files on both of them, then you would use Windows filenames ("C:\foo\bar.txt") for the files on the Windows machine and UNIX file names (/usr/local/src/foo/bar") to refer to the UNIX file names. If you were running a program (perhaps merely the command shell), and you wanted to refer to two different file systems (maybe a Copy command, or maybe some of the application data was on one server and some was on the other), then you could type both file names in the manner that was native for each of the different file systems. The operating system understood that these were filenames, as opposed to some arbitrary string pattern. So it could do all kinds of fancy things for you (based on having parsed the file names, and having all kinds of knowledge about how files and filenames worked on the particular file server's operating system.) In the case of UNIX, that might not be clicking with you because filenames don't have types and things, so that might not mean too much to you since UNIX does not have not have structured file names -- they're just strings. But people on UNIX often use certain conventions (such as ".c" means it's a C source file) and they use globbing in certain ways in order to mimic such effects. So the system knew about those conventions and would emulate them for you. On most operating systems, filenames have different components. Maybe there is a standard "type" part and a "version" part. There might be some known file types, such as TXT and BAT and EXE. On some other system, the text file type might be "text" instead of "TXT". Suppose I wish to copy a text file from my UNIX machine to my DOS machine. I could type :Copy File U:/foo/foobar.text D:*\*.* and on the "D" machine (running DOS) the resulting file would be named "D:\FOO\FOOBAR.TXT", because the system knew about that. (You could override that sort of behaviour if you wanted.) If you were in the middle of typing a filename on a filesystem that supported version numbers (like ">CStacy>work>hack>foo.cpp.69"), then when I got to the part where that "[0-9]*\|newest|oldest" part went, then it would beep at me if I tried to type something other than numeric digits or the special tokens "oldest" and "newest". Finally, if you didn't like using the native syntax, you could use a "logical" syntax that mapped the native format into one generic format, instead. Such a logical host filename could refer to more than one actual file server, so that some of the directories lived in one place, and some elsewhere, even across different kinds of filesystems. (I might keep some sources on my UNIX host and font library files on my Macintosh, and refer to them as though they were simply different directories on a "logical" host.) Notice that *all* applications get these features, not just shell commands. This is different than on UNIX, where, for example, you only get globbing on file names that typed into the shell. So if you have a command-line based program, you can pass in some wildcarded filenames. But then a window pops up and asks the user for some file names. You don't get globbing there, because it's not a standard part of the system. Even if suppose it was available in a library, sometimes the programmer would do it and sometimes not. But on operating systems like the Lisp Machine, you always get it, everywhere (unless for some strange reason the programmer wanted to deliberately prevent it from working). ###### Newsgroups: alt.sys.pdp8,alt.sys.pdp10,alt.folklore.computers Subject: Re: Filenames, shells, remote file access [was: ooh, a *real* flamewar :) References:

<8bg02p$1b2k$1@citadel.in.taronga.com> Organization: Plethora . Net - More Net, Less Spam! X-Newsreader: trn 4.0-test72 (19 April 1999) From: seebs@plethora.net (Peter Seebach) Lines: 63 Message-ID: Date: Fri, 24 Mar 2000 23:54:56 GMT NNTP-Posting-Host: 205.166.146.8 X-Complaints-To: abuse@plethora.net X-Trace: ptah.visi.com 953942096 205.166.146.8 (Fri, 24 Mar 2000 17:54:56 CST) NNTP-Posting-Date: Fri, 24 Mar 2000 17:54:56 CST Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news.maxwell.syr.edu!hermes.visi.com!news-out.visi.com!ptah.visi.com!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:52850 In article , Christopher C Stacy wrote: >The point is that the operating system allowed users to keep their >files on whatever kind of remote filesystem they preferred, and refer >to them in the way that would be most familiar to the user. I don't find that convincing at all. Among other things, it obviously had its own local convention (initial "box:" indicates a host). Did it ever enforce any other conventions on programs? At all? Could you have null bytes in filenames? I would rather see a system provide a *translation* of remote names. There's a famous story about a system which did precisely what you describe; once upon a time, there was an NFS server which spoke to macs, and it simply passed filenames back and forth, resulting in files with names containing '/' characters, on Unix boxes... ;-) >If you had several different kinds of file servers in the shop, >and you did some of your development on Windows and some on UNIX >and kept files on both of them, then you would use Windows filenames >("C:\foo\bar.txt") for the files on the Windows machine and UNIX file >names (/usr/local/src/foo/bar") to refer to the UNIX file names. Neat, so it's *TOTALLY* impossible for you to interact with a file without knowing what type of computer it came from. How thoughtful! >If you were running a program (perhaps merely the command shell), >and you wanted to refer to two different file systems (maybe a Copy >command, or maybe some of the application data was on one server and >some was on the other), then you could type both file names in the >manner that was native for each of the different file systems. And if you didn't, you were screwed. >Finally, if you didn't like using the native syntax, you could use a "logical" >syntax that mapped the native format into one generic format, instead. That would make a bit more sense... >But then a window pops up and asks the user for some file names. >You don't get globbing there, because it's not a standard part of the system. You do if the programmer wanted to provide it... But what does the system that *ALWAYS* globs do if a program doesn't want globbing? >Even if suppose it was available in a library, sometimes the programmer would >do it and sometimes not. But on operating systems like the Lisp Machine, >you always get it, everywhere (unless for some strange reason the programmer >wanted to deliberately prevent it from working). I don't see this as a huge win either way. Mostly, I see it as a fairly psychotic decision about how to handle heterogeneous networks. Neat, but not something I'd want to use. -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/ Get paid to surf! No spam. http://www.alladvantage.com/go.asp?refid=GZX636 ###### From: peter@taronga.com (Peter da Silva) Newsgroups: alt.sys.pdp8,alt.sys.pdp10,alt.folklore.computers Subject: Re: Filenames, shells, remote file access [was: ooh, a *real* flamewar :) Date: 25 Mar 2000 00:53:09 GMT Organization: TSS Inc. Lines: 79 Message-ID: <8bh2ll$21lo$1@citadel.in.taronga.com> References:

<8bg02p$1b2k$1@citadel.in.taronga.com> NNTP-Posting-Host: citadel.in.taronga.com X-Trace: citadel.in.taronga.com 953945589 67256 10.0.0.43 (25 Mar 2000 00:53:09 GMT) X-Complaints-To: usenet@taronga.com NNTP-Posting-Date: 25 Mar 2000 00:53:09 GMT X-Newsreader: trn 4.0-test72 (19 April 1999) 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!news-proxy.baileynm.com!citadel.in.taronga.com!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:52875 In article , Christopher C Stacy wrote: >The point is that the operating system allowed users to keep their >files on whatever kind of remote filesystem they preferred, and refer >to them in the way that would be most familiar to the user. But only from the Lisp Machine environment. You couldn't use this to access VMS files from the DOS environment, like you could on OpenNET. Since you're working in a completely different environment, then details of file names are the least of your worries... >If they had a bunch of UNIX boxes that they liked, then they could >just access those files using the regular UNIX file names. I had Xenix, System V, VMS, and DOS boxes. People on DOS could access resources on any system using their familiar DOS names. People on Xenix could do the same things using their familiar Xenix names. VMS wasn't quite that transparent, alas. You had to use a special command shell (like FTP), so it didn't bother translating the names to VMS syntax. There are advantages to both ways of doing things. I prefer the symmetry of OpenNET. >The operating system understood that these were filenames, as opposed >to some arbitrary string pattern. So it could do all kinds of fancy >things for you (based on having parsed the file names, and having >all kinds of knowledge about how files and filenames worked on the >particular file server's operating system.) But only on the operating system that did that for you. On OpenNET you could largely make the network invisible. You didn't have to know what type of system a remote resource was on. >In the case of UNIX, that might not be clicking with you because >filenames don't have types and things, so that might not mean too much >to you since UNIX does not have not have structured file names -- I've possibly used more different kinds of file systems than most people in this thread, including the real old-timers like BAH. OS/1100 makes anything else look like UNIX compared to its baroqueness: qualifier*filename/readkey/writekey(f-cycle).elt/ver(elt-cycle) They later added a machine-name as well, but I've mercifully recycled the neurons that remembered the syntax. I think it might have been: machine#qualifier*filename/readkey/writekey(f-cycle).elt/ver(elt-cycle) Elements were implemented as archive members in files, but used the way files were used in other systems. You could have multiple versions (cycles) of a given file (element), and four types (versions) of a file. Plus the directory (file) itself had a version (f-cycle), read and write keys, and a qualifier (sort of like the user part in an RSX-11 UIC). This is the only file name syntax that I've run into that I didn't try to handle in my library. I hope you all can forgive me. >You don't get globbing there, because it's not a standard part of the system. >Even if suppose it was available in a library, sometimes the programmer would >do it and sometimes not. But on operating systems like the Lisp Machine, >you always get it, everywhere (unless for some strange reason the programmer >wanted to deliberately prevent it from working). Well, that's all great. I've heard lots of good things about the lisp machines, and I'm quite willing to believe that they were the closest thing to perfect we've yet seen, I was just wondering, though, how they dealt with recognising all the different syntaxes... what would happen if you tried to use a file name on a machine that wasn't currently accessible, but you needed to twiddle for some reason (say putting it into some command you were going to kick off at a scheduled time when the resource would be available). As I said, even dealing with it one syntax at a time was painful. -- This is The Reverend Peter da Silva's Boring Sig File - there are no references to Wolves, Kibo, Discordianism, or The Church of the Subgenius in this document Executive Vice President, Corporate Communications, Entropy Gradient Reversals. ###### From: jmfbahciv@aol.com Newsgroups: alt.sys.pdp8,alt.sys.pdp10,alt.folklore.computers Subject: Re: Filenames, shells, remote file access [was: ooh, a *real* flamewar :) Date: Sun, 26 Mar 00 11:18:59 GMT Organization: UltraNet Communications, Inc. Lines: 74 Message-ID: <8bl1dj$vv$3@bob.news.rcn.net> References: <38da49e6_2@news.wizvax.net> <8beb7i$707$1@citadel.in.taronga.com> <8bg02p$1b2k$1@citadel.in.taronga.com> X-Trace: 9IGZVfLHro9gCMiLtHnRlbV4CZAC9MoPdcOJhf03gzc= X-Complaints-To: abuse@rcn.com NNTP-Posting-Date: 26 Mar 2000 12:56:19 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!skynet.be!triton.skycache.com!europa.netcrusader.net!207.172.3.37!feed1.news.rcn.net!rcn!207-172-216-186 Xref: chonsp.franklin.ch alt.folklore.computers:52926 In article <8bg02p$1b2k$1@citadel.in.taronga.com>, peter@taronga.com (Peter da Silva) wrote: >In article , >Christopher C Stacy wrote: >>The file name was not mapped to a standard convention - you used the remote >>file system's syntax. So if you wanted a file "/home/cstacy/foobar" that >>lived on some Unix box named BART, the file "Open" library call would take >>this string: >> >> BART:/home/cstacy/Foobar.txt >> >>For a similarly named file on TOPS-20, it would be >> >> LISA:FOOBAR.TXT > >I'm somewhat boggled. > >It's hard enough dealing with one file naming convention at a time... I've >written libraries to do that, to make it easier to write >programs that would >run on UNIX, AmigaOS, VMS, and DOS. Trying to write a program that did the >right thing with a list of files that might have >completely different syntaxes >seems impossible. How did it deal with this? The only think I can think of >that would work would be to know what kind of system each was... and that >would break down if you had to deal with file names on >machines that weren't >currently available on the network. No. The way the specs read was that the character string containing the full file spec typed in the command was passed to the target computer. That computer was the one who then parsed the ASCII and converted it into a file spec that its OS knew and loved dearly. Once the bits were converted to a file spec, then the operation requested was serviced. If it was a file copy, it just passed the _contents_ of the file it found back to the requestor who knew how to deal with its OS. > >Or did you simply not do that? > >By the way, there were a couple of early UNIX remote file systems that came >much closer to what you're talking about. Rather than step through the remote >file system step by step, the "remote" part of a file name was simply passed >to the other and for translation. File systems on the network used a standard >syntax, though, so when you looked at them from a DOS box they had DOS syntax >so DOS programs didn't think UNIX filenames were command options, and UNIX >shells saw UNIX paths so they didn't treat VMS directories as regular >expressions. > >If you're trying to make a system that works transparently from all sides, in >a heterogenous environment, you can't just optimise for one operating system. > But one doesn't really have anything to do with the other. I think one of the confusions happening here is that you're don't know which level of software is dealing with a piece of functionality. /BAH Subtract a hundred and four for e-mail. ###### From: peter@taronga.com (Peter da Silva) Newsgroups: alt.sys.pdp8,alt.sys.pdp10,alt.folklore.computers Subject: Re: Filenames, shells, remote file access [was: ooh, a *real* flamewar :) Date: 26 Mar 2000 15:30:51 GMT Organization: TSS Inc. Lines: 15 Message-ID: <8blafb$tm3$1@citadel.in.taronga.com> References:

<8bg02p$1b2k$1@citadel.in.taronga.com> <8bl1dj$vv$3@bob.news.rcn.net> NNTP-Posting-Host: citadel.in.taronga.com X-Trace: citadel.in.taronga.com 954084651 30403 10.0.0.43 (26 Mar 2000 15:30:51 GMT) X-Complaints-To: usenet@taronga.com NNTP-Posting-Date: 26 Mar 2000 15:30:51 GMT X-Newsreader: trn 4.0-test72 (19 April 1999) Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news-fra1.dfn.de!news0.de.colt.net!colt.net!news.maxwell.syr.edu!newsfeed.stanford.edu!news-proxy.baileynm.com!citadel.in.taronga.com!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:52953 In article <8bl1dj$vv$3@bob.news.rcn.net>, wrote: >No. The way the specs read was that the character string >containing the full file spec typed in the command >was passed to the target computer. That computer was the >one who then parsed the ASCII and converted it into a >file spec that its OS knew and loved dearly. So programs that were expecting to be able to take a file name and understand it were pretty much impossible. Which was my point. -- This is The Reverend Peter da Silva's Boring Sig File - there are no references to Wolves, Kibo, Discordianism, or The Church of the Subgenius in this document Executive Vice President, Corporate Communications, Entropy Gradient Reversals. ###### From: jmfbahciv@aol.com Newsgroups: alt.sys.pdp8,alt.sys.pdp10,alt.folklore.computers Subject: Re: Filenames, shells, remote file access [was: ooh, a *real* flamewar :) Date: Tue, 28 Mar 00 09:26:30 GMT Organization: UltraNet Communications, Inc. Lines: 39 Message-ID: <8bq3j8$nqs$2@bob.news.rcn.net> References:

<8bg02p$1b2k$1@citadel.in.taronga.com> <8bl1dj$vv$3@bob.news.rcn.net> <8blafb$tm3$1@citadel.in.taronga.com> X-Trace: DnPEscmxQcnh4hcA/arK/BbkF+AofipKeEPn8EJwwLo= X-Complaints-To: abuse@rcn.com NNTP-Posting-Date: 28 Mar 2000 11:04:08 GMT X-Newsreader: News Xpress Version 1.0 Beta #4 Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news.maxwell.syr.edu!feed1.news.rcn.net!rcn!207-172-245-49 Xref: chonsp.franklin.ch alt.folklore.computers:53108 In article <8blafb$tm3$1@citadel.in.taronga.com>, peter@taronga.com (Peter da Silva) wrote: >In article <8bl1dj$vv$3@bob.news.rcn.net>, wrote: >>No. The way the specs read was that the character string >>containing the full file spec typed in the command >>was passed to the target computer. That computer was the >>one who then parsed the ASCII and converted it into a >>file spec that its OS knew and loved dearly. > >So programs that were expecting to be able to take a >file name and understand >it were pretty much impossible. Which was my point. > Huh? Look, when an application is written it is part of the spec of where the demunging happens. If my program on a host computer always knows what the target operating system is going to be, then I can simply pass the demunged argument list to get at a file (after all, I should know exactly what I want to read). However, if I'm writing a program that is to implement a host to host file copy mechanism, then that program shouldn't make any assumptions about what the target file specification format is supposed to be. And the reason for that is so that the copy program doesn't have to be modified just because some other group or company changes its file specification format. The way things look really depends on the function of the software that is being used. There isn't just one glorious way to access data on another system that's on the network. One program's function is just copy what the user wants (and this is never going to be predictable) and another program's function is retrieve record foo from data base x on system z. /BAH /BAH Subtract a hundred and four for e-mail. ###### From: peter@taronga.com (Peter da Silva) Newsgroups: alt.sys.pdp8,alt.sys.pdp10,alt.folklore.computers Subject: Re: Filenames, shells, remote file access [was: ooh, a *real* flamewar :) Date: 28 Mar 2000 16:27:03 GMT Organization: TSS Inc. Lines: 30 Message-ID: <8bqmgn$1pas$1@citadel.in.taronga.com> References: <8bl1dj$vv$3@bob.news.rcn.net> <8blafb$tm3$1@citadel.in.taronga.com> <8bq3j8$nqs$2@bob.news.rcn.net> NNTP-Posting-Host: citadel.in.taronga.com X-Trace: citadel.in.taronga.com 954260823 58716 10.0.0.43 (28 Mar 2000 16:27:03 GMT) X-Complaints-To: usenet@taronga.com NNTP-Posting-Date: 28 Mar 2000 16:27:03 GMT X-Newsreader: trn 4.0-test72 (19 April 1999) Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!News.Amsterdam.UnisourceCS!skynet.be!newsfeed.stanford.edu!news-proxy.baileynm.com!citadel.in.taronga.com!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:53146 In article <8bq3j8$nqs$2@bob.news.rcn.net>, wrote: >In article <8blafb$tm3$1@citadel.in.taronga.com>, > peter@taronga.com (Peter da Silva) wrote: >>So programs that were expecting to be able to take a >>file name and understand >>it were pretty much impossible. Which was my point. >Huh? Look, when an application is written it is part of the >spec of where the demunging happens. If my program on a host >computer always knows what the target operating system is >going to be, then I can simply pass the demunged argument list >to get at a file (after all, I should know exactly what I >want to read). Here's a filename. Your program is a compiler. It's going to want to put the output of the compile in a file with the extension "OBJ". The source file is in COBOL: VMS:DUA0:[peter.cobol.work]FRED.COB;1 MULTICS:>Users>Peter>COBOL>Work>Joe.Cob.1 EXEC1100:peter*cobolwork.mikesource/sym(1) What's a poor program to do? How does it even know what's the file name and what's the extension? -- This is The Reverend Peter da Silva's Boring Sig File - there are no references to Wolves, Kibo, Discordianism, or The Church of the Subgenius in this document Executive Vice President, Corporate Communications, Entropy Gradient Reversals. ###### From: jmaynard@thebrain.conmicro.cx (Jay Maynard) Newsgroups: alt.sys.pdp8,alt.sys.pdp10,alt.folklore.computers Subject: Re: Filenames, shells, remote file access [was: ooh, a *real* flamewar :) References: <8bl1dj$vv$3@bob.news.rcn.net> <8blafb$tm3$1@citadel.in.taronga.com> <8bq3j8$nqs$2@bob.news.rcn.net> <8bqmgn$1pas$1@citadel.in.taronga.com> Reply-To: jmaynard@texas.net Message-ID: User-Agent: slrn/0.9.5.4 (UNIX) Lines: 16 NNTP-Posting-Date: Tue, 28 Mar 2000 11:34:52 CST Organization: Giganews.Com - Premium News Outsourcing X-Trace: sv2-ekz4V1N4M62PNFh2nyfxblS6cgzEwGzFHFcyMiHC5EOLcJOMDs4Fti13UVUERZWmSIPg0R104aQKoz/!8CBcPTk5L7aNiBpXQKm4hQ== X-Complaints-To: abuse@GigaNews.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 Date: Tue, 28 Mar 2000 17:34:52 GMT Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!isdnet!netnews.com!feed1.news.rcn.net!rcn!nntp.giganews.com!nntp3.giganews.com!news5.giganews.com.POSTED!jmaynard Xref: chonsp.franklin.ch alt.folklore.computers:53084 On 28 Mar 2000 16:27:03 GMT, Peter da Silva wrote: >Here's a filename. Your program is a compiler. It's going to want to put >the output of the compile in a file with the extension "OBJ". The >source file is in COBOL: >VMS:DUA0:[peter.cobol.work]FRED.COB;1 >MULTICS:>Users>Peter>COBOL>Work>Joe.Cob.1 >EXEC1100:peter*cobolwork.mikesource/sym(1) Let me toss in another one: MVS:PETER.APPLIC8N.COBOL(HARRY) >What's a poor program to do? How does it even know what's the file name and >what's the extension? Especially when there are few clues you can glean from delimiters? ###### From: jmfbahciv@aol.com Newsgroups: alt.sys.pdp8,alt.sys.pdp10,alt.folklore.computers Subject: Re: Filenames, shells, remote file access [was: ooh, a *real* flamewar :) Date: Wed, 29 Mar 00 11:31:20 GMT Organization: UltraNet Communications, Inc. Lines: 50 Message-ID: <8bsv9i$9kv$2@bob.news.rcn.net> References: <8bl1dj$vv$3@bob.news.rcn.net> <8blafb$tm3$1@citadel.in.taronga.com> <8bq3j8$nqs$2@bob.news.rcn.net> <8bqmgn$1pas$1@citadel.in.taronga.com> X-Trace: 6p+so5bkspZDeikWLM0HXuCwp06UN2kBmqHyVcF7roc= X-Complaints-To: abuse@rcn.com NNTP-Posting-Date: 29 Mar 2000 13:09:06 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!207-172-102-231 Xref: chonsp.franklin.ch alt.folklore.computers:53201 In article , jmaynard@thebrain.conmicro.cx (Jay Maynard) wrote: >On 28 Mar 2000 16:27:03 GMT, Peter da Silva wrote: >>Here's a filename. Your program is a compiler. It's going to want to put >>the output of the compile in a file with the extension "OBJ". The >>source file is in COBOL: >>VMS:DUA0:[peter.cobol.work]FRED.COB;1 >>MULTICS:>Users>Peter>COBOL>Work>Joe.Cob.1 >>EXEC1100:peter*cobolwork.mikesource/sym(1) > >Let me toss in another one: > >MVS:PETER.APPLIC8N.COBOL(HARRY) > >>What's a poor program to do? How does it even know what's >>the file name and >>what's the extension? > >Especially when there are few clues you can glean from delimiters? That's nonsense. There's all kinds of clues. The system where you logged in (I'm going to use TOPS10 cybercrud here) takes your command and parses it. It sees that you are asking for a file that is not on the system's local disk file system. Now it can make an assumption that the file exists on another system and has to make the assumption that the target system is available on the network and able to handle network file specs. So the text of your command to compile is shipped over to the target system where the file actually exists. Now, do you want the compilation to happen over there? Or do you want the compilation to happen over here? The way we distinguished that desire was through the initial command (this is determined by the functional specification of the command interpreter). One scenario is that the file you want to compile is first copied to your host computer and then compiled. All that need to be done is to ship the file spec that the user typed over to the target computer and ask for the target computer to take the file spec, find the file and ship it back to your host computer. Then normal compilation steps are taken to do the compilation. IT'S A USER'S DECISION to tell the computer what he wants done, where he wants it done, and where he wants the output to go. /BAH Subtract a hundred and four for e-mail. ###### From: jmfbahciv@aol.com Newsgroups: alt.sys.pdp8,alt.sys.pdp10,alt.folklore.computers Subject: Re: Filenames, shells, remote file access [was: ooh, a *real* flamewar :) Date: Wed, 29 Mar 00 11:23:05 GMT Organization: UltraNet Communications, Inc. Lines: 37 Message-ID: <8bsuq3$9kv$1@bob.news.rcn.net> References: <8bl1dj$vv$3@bob.news.rcn.net> <8blafb$tm3$1@citadel.in.taronga.com> <8bq3j8$nqs$2@bob.news.rcn.net> <8bqmgn$1pas$1@citadel.in.taronga.com> X-Trace: bSxi+SS8V0RkCyhGJKEcp51Cfp9gZgE+5W5V521pO94= X-Complaints-To: abuse@rcn.com NNTP-Posting-Date: 29 Mar 2000 13:00:51 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!207-172-102-231 Xref: chonsp.franklin.ch alt.folklore.computers:53202 In article <8bqmgn$1pas$1@citadel.in.taronga.com>, peter@taronga.com (Peter da Silva) wrote: >In article <8bq3j8$nqs$2@bob.news.rcn.net>, wrote: >>In article <8blafb$tm3$1@citadel.in.taronga.com>, >> peter@taronga.com (Peter da Silva) wrote: >>>So programs that were expecting to be able to take a >>>file name and understand >>>it were pretty much impossible. Which was my point. > >>Huh? Look, when an application is written it is part of the >>spec of where the demunging happens. If my program on a host >>computer always knows what the target operating system is >>going to be, then I can simply pass the demunged argument list >>to get at a file (after all, I should know exactly what I >>want to read). > >Here's a filename. Your program is a compiler. It's going to want to put >the output of the compile in a file with the extension "OBJ". The >source file is in COBOL: > >VMS:DUA0:[peter.cobol.work]FRED.COB;1 >MULTICS:>Users>Peter>COBOL>Work>Joe.Cob.1 >EXEC1100:peter*cobolwork.mikesource/sym(1) > >What's a poor program to do? How does it even know what's >the file name and what's the extension? > The compiler doesn't need to know the full file spec of the set of ASCII characters that needs to be compiled; the user knows. It is a user command to the compiler that tells the compiler what to compile and where to get the code to compile. /BAH Subtract a hundred and four for e-mail. ###### From: jmfbahciv@aol.com Newsgroups: alt.sys.pdp8,alt.sys.pdp10,alt.folklore.computers Subject: Re: Filenames, shells, remote file access [was: ooh, a *real* flamewar :) Date: Thu, 30 Mar 00 09:12:55 GMT Organization: UltraNet Communications, Inc. Lines: 69 Message-ID: <8bvbie$20t$1@bob.news.rcn.net> References: <8bqmgn$1pas$1@citadel.in.taronga.com> <8bsv9i$9kv$2@bob.news.rcn.net> <8bthc5$2d2i$1@citadel.in.taronga.com> X-Trace: yejMtUEQLVWQnjbQiZipya+w4WSNTp90h5pt1Lwrupw= X-Complaints-To: abuse@rcn.com NNTP-Posting-Date: 30 Mar 2000 10:50:54 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!207-172-102-24 Xref: chonsp.franklin.ch alt.folklore.computers:53322 In article <8bthc5$2d2i$1@citadel.in.taronga.com>, peter@taronga.com (Peter da Silva) wrote: >In article <8bsv9i$9kv$2@bob.news.rcn.net>, wrote: >>In article , >> jmaynard@thebrain.conmicro.cx (Jay Maynard) wrote: >>>On 28 Mar 2000 16:27:03 GMT, Peter da Silva wrote: >>>>Here's a filename. Your program is a compiler. It's going to want to put >>>>the output of the compile in a file with the extension "OBJ". The >>>>source file is in COBOL: >>>>VMS:DUA0:[peter.cobol.work]FRED.COB;1 >>>>MULTICS:>Users>Peter>COBOL>Work>Joe.Cob.1 >>>>EXEC1100:peter*cobolwork.mikesource/sym(1) > >>>Let me toss in another one: > >>>MVS:PETER.APPLIC8N.COBOL(HARRY) > >>>>What's a poor program to do? How does it even know what's >>>>the file name and >>>>what's the extension? > >>>Especially when there are few clues you can glean from delimiters? > >>That's nonsense. There's all kinds of clues. The system where >>you logged in (I'm going to use TOPS10 cybercrud here) takes >>your command and parses it. > >The system I'm logged in at hasn't seen the file yet. It's >the cobol compiler that has it, and it's trying to generate >the output file name so it can ask the OS to open it. How did the Cobol compiler get initiated to start the compile in the first place? It had to be some flavor of a user command. > >>One scenario is that the file you want to compile is first >>copied to your host computer and then compiled. All that >>need to be done is to ship the file spec that the user typed >>over to the target computer and ask for the target computer >>to take the file spec, find the file and ship it back to >>your host computer. > >Um, that's not what I think of as network file access. Why >should it have to ship a file over here before I even start >to read it? Hell, I may not have even opened it yet... Reading any file that is non-local dictates that the file gets shipped to the local host...always. > >Or do you mean that every time I run the COBOL compiler I have to tell it >the destination file name as well as the source file name? Of course you do. Just because you don't explicitly give a file destination doesn't mean that you haven't implicitly specified it...we used call that filling in the defaults. > >I would personally consider that user-hostile. Computers don't read minds but they do make assumptions when necessary data isn't specified by the users. User hostility is when that data is filled in without giving the user the option. /BAH Subtract a hundred and four for e-mail. ###### From: dg@pearl.tao.co.uk (David Given) Newsgroups: alt.sys.pdp8,alt.sys.pdp10,alt.folklore.computers Subject: Re: Filenames, shells, remote file access [was: ooh, a *real* flamewar :) Date: Wed, 29 Mar 2000 14:22:16 +0100 Organization: I do not speak for anyone but myself, and barely that. Lines: 28 Message-ID: <820tb8.qc7.ln@127.0.0.1> References: <8bl1dj$vv$3@bob.news.rcn.net> <8blafb$tm3$1@citadel.in.taronga.com> <8bq3j8$nqs$2@bob.news.rcn.net> <8bqmgn$1pas$1@citadel.in.taronga.com> NNTP-Posting-Host: 212.19.67.123 Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Trace: 954357744 IIX5YQT0T437BD413C uk21.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!news-ge.switch.ch!news.maxwell.syr.edu!newsfeed.berkeley.edu!sn-xit-01!supernews.com!sn-inject-01!127.0.0.1!nobody Xref: chonsp.franklin.ch alt.folklore.computers:53394 In article , jmaynard@thebrain.conmicro.cx (Jay Maynard) writes: > On 28 Mar 2000 16:27:03 GMT, Peter da Silva wrote: >>Here's a filename. Your program is a compiler. It's going to want to put >>the output of the compile in a file with the extension "OBJ". The >>source file is in COBOL: >>VMS:DUA0:[peter.cobol.work]FRED.COB;1 >>MULTICS:>Users>Peter>COBOL>Work>Joe.Cob.1 >>EXEC1100:peter*cobolwork.mikesource/sym(1) > > Let me toss in another one: > > MVS:PETER.APPLIC8N.COBOL(HARRY) This is fun! RISCOS:adfs::0.Users.Peter.Cobol.Work.Fred Add to this the fact that Risc OS doesn't use file extensions (it uses file types instead, three-digit hex codes that describe each file. You can't have two files in the same directory with the same name but different types). -- +- David Given ---------------McQ-+ | Work: dg@tao-group.com | Look behind you! A three-headed monkey! | Play: dgiven@iname.com | +- http://wired.st-and.ac.uk/~dg -+ ###### From: peter@taronga.com (Peter da Silva) Newsgroups: alt.sys.pdp8,alt.sys.pdp10,alt.folklore.computers Subject: Re: Filenames, shells, remote file access [was: ooh, a *real* flamewar :) Date: 29 Mar 2000 18:17:41 GMT Organization: TSS Inc. Lines: 50 Message-ID: <8bthc5$2d2i$1@citadel.in.taronga.com> References: <8bqmgn$1pas$1@citadel.in.taronga.com> <8bsv9i$9kv$2@bob.news.rcn.net> NNTP-Posting-Host: citadel.in.taronga.com X-Trace: citadel.in.taronga.com 954353861 78930 10.0.0.43 (29 Mar 2000 18:17:41 GMT) X-Complaints-To: usenet@taronga.com NNTP-Posting-Date: 29 Mar 2000 18:17:41 GMT X-Newsreader: trn 4.0-test72 (19 April 1999) Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!isdnet!logbridge.uoregon.edu!newsfeed.stanford.edu!news-proxy.baileynm.com!citadel.in.taronga.com!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:53366 In article <8bsv9i$9kv$2@bob.news.rcn.net>, wrote: >In article , > jmaynard@thebrain.conmicro.cx (Jay Maynard) wrote: >>On 28 Mar 2000 16:27:03 GMT, Peter da Silva wrote: >>>Here's a filename. Your program is a compiler. It's going to want to put >>>the output of the compile in a file with the extension "OBJ". The >>>source file is in COBOL: >>>VMS:DUA0:[peter.cobol.work]FRED.COB;1 >>>MULTICS:>Users>Peter>COBOL>Work>Joe.Cob.1 >>>EXEC1100:peter*cobolwork.mikesource/sym(1) >>Let me toss in another one: >>MVS:PETER.APPLIC8N.COBOL(HARRY) >>>What's a poor program to do? How does it even know what's >>>the file name and >>>what's the extension? >>Especially when there are few clues you can glean from delimiters? >That's nonsense. There's all kinds of clues. The system where >you logged in (I'm going to use TOPS10 cybercrud here) takes >your command and parses it. The system I'm logged in at hasn't seen the file yet. It's the cobol compiler that has it, and it's trying to generate the output file name so it can ask the OS to open it. >One scenario is that the file you want to compile is first >copied to your host computer and then compiled. All that >need to be done is to ship the file spec that the user typed >over to the target computer and ask for the target computer >to take the file spec, find the file and ship it back to >your host computer. Um, that's not what I think of as network file access. Why should it have to ship a file over here before I even start to read it? Hell, I may not have even opened it yet... Or do you mean that every time I run the COBOL compiler I have to tell it the destination file name as well as the source file name? I would personally consider that user-hostile. -- This is The Reverend Peter da Silva's Boring Sig File - there are no references to Wolves, Kibo, Discordianism, or The Church of the Subgenius in this document Executive Vice President, Corporate Communications, Entropy Gradient Reversals. ###### From: peter@taronga.com (Peter da Silva) Newsgroups: alt.sys.pdp8,alt.sys.pdp10,alt.folklore.computers Subject: Re: Filenames, shells, remote file access [was: ooh, a *real* flamewar :) Date: 29 Mar 2000 18:29:17 GMT Organization: TSS Inc. Lines: 58 Message-ID: <8bti1t$2dcq$1@citadel.in.taronga.com> References: <8bq3j8$nqs$2@bob.news.rcn.net> <8bqmgn$1pas$1@citadel.in.taronga.com> <8bsuq3$9kv$1@bob.news.rcn.net> NNTP-Posting-Host: citadel.in.taronga.com X-Trace: citadel.in.taronga.com 954354557 79258 10.0.0.43 (29 Mar 2000 18:29:17 GMT) X-Complaints-To: usenet@taronga.com NNTP-Posting-Date: 29 Mar 2000 18:29:17 GMT X-Newsreader: trn 4.0-test72 (19 April 1999) Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!isdnet!logbridge.uoregon.edu!newsfeed.stanford.edu!news-proxy.baileynm.com!citadel.in.taronga.com!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:53374 In article <8bsuq3$9kv$1@bob.news.rcn.net>, wrote: >The compiler doesn't need to know the full file spec of >the set of ASCII characters that needs to be compiled; >the user knows. It is a user command to the compiler >that tells the compiler what to compile and where to get >the code to compile. So every time I want to compile a program I have to tell the compiler both the source name and the destination name and the names for it to use for intermediate object files? >>VMS:DUA0:[peter.cobol.work]FRED.COB;1 I used to do this: % cd //vms/peter/cobol/work % ls fred.cob % cob fred.com % ls fred.cob fred.obj % ld fred.obj -lcobol % ls cob.out fred.cob fred.obj Meanwhile if you logged into the VMS box you see: $ SET DEFAULT DUA0:[peter.cobol.work] $ DIR FRED.COB;1 FRED.OBJ;1 COB.OUT;1 Or on a DOS box you could do: C:\> net use x: \\vms\peter C:\> x: X:\> cd cobol X:\COBOL\> dir/w fred.cob fred.obj cob.out 3 file(s) 24,509 bytes 312,228,480 bytes free You literally didn't need to know what operating system a file was on. It was completely transparent on all machines that had the OpenNET client. Very cool. Much nicer than having to keep track of filename syntax for a bunch of machines you didn't know about. -- This is The Reverend Peter da Silva's Boring Sig File - there are no references to Wolves, Kibo, Discordianism, or The Church of the Subgenius in this document Executive Vice President, Corporate Communications, Entropy Gradient Reversals. ###### From: peter@taronga.com (Peter da Silva) Newsgroups: alt.sys.pdp8,alt.sys.pdp10,alt.folklore.computers Subject: Re: Filenames, shells, remote file access [was: ooh, a *real* flamewar :) Date: 30 Mar 2000 14:35:50 GMT Organization: TSS Inc. Lines: 38 Message-ID: <8bvoo7$r6i$1@citadel.in.taronga.com> References: <8bsv9i$9kv$2@bob.news.rcn.net> <8bthc5$2d2i$1@citadel.in.taronga.com> <8bvbie$20t$1@bob.news.rcn.net> NNTP-Posting-Host: citadel.in.taronga.com X-Trace: citadel.in.taronga.com 954426951 27858 10.0.0.43 (30 Mar 2000 14:35:50 GMT) X-Complaints-To: usenet@taronga.com NNTP-Posting-Date: 30 Mar 2000 14:35:50 GMT X-Newsreader: trn 4.0-test72 (19 April 1999) Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!isdnet!logbridge.uoregon.edu!newsfeed.stanford.edu!news-proxy.baileynm.com!citadel.in.taronga.com!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:53363 In article <8bvbie$20t$1@bob.news.rcn.net>, wrote: >>Um, that's not what I think of as network file access. Why >>should it have to ship a file over here before I even start >>to read it? Hell, I may not have even opened it yet... >Reading any file that is non-local dictates that the file >gets shipped to the local host...always. Why? That's what the VICE file system in Andrew did, and I believe AFS still does this, and I suppose for most cases that's good enough. It really breaks things when you're doing something like looking at the last few lines of an 11 meg logfile or grabbing a couple of records from a database. And in any case this should be handled behind the program's back. To the user it looks like he's directly accessing the file on the remote box. >>Or do you mean that every time I run the COBOL compiler I have to tell it >>the destination file name as well as the source file name? >Of course you do. Just because you don't explicitly give >a file destination doesn't mean that you haven't implicitly >specified it...we used call that filling in the defaults. OK, how does the COBOL compiler figure out what the defaults should be for these? >VMS:DUA0:[peter.cobol.work]FRED.COB;1 >MULTICS:>Users>Peter>COBOL>Work>Joe.Cob.1 >EXEC1100:peter*cobolwork.mikesource/sym(1) >MVS:PETER.APPLIC8N.COBOL(HARRY) -- This is The Reverend Peter da Silva's Boring Sig File - there are no references to Wolves, Kibo, Discordianism, or The Church of the Subgenius in this document Executive Vice President, Corporate Communications, Entropy Gradient Reversals. ###### From: jmfbahciv@aol.com Newsgroups: alt.sys.pdp8,alt.sys.pdp10,alt.folklore.computers Subject: Re: Filenames, shells, remote file access [was: ooh, a *real* flamewar :) Date: Fri, 31 Mar 00 09:19:19 GMT Organization: UltraNet Communications, Inc. Lines: 25 Message-ID: <8c20an$npb$1@bob.news.rcn.net> References: <8bqmgn$1pas$1@citadel.in.taronga.com> <8bsv9i$9kv$2@bob.news.rcn.net> <8bthc5$2d2i$1@citadel.in.taronga.com> <8bvbie$20t$1@bob.news.rcn.net> X-Trace: LCSQ3nkwRkZNZm/IC1WJKFyptbJozo6+LbCgBka2DqU= X-Complaints-To: abuse@rcn.com NNTP-Posting-Date: 31 Mar 2000 10:57:27 GMT X-Newsreader: News Xpress Version 1.0 Beta #4 Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news.maxwell.syr.edu!feed1.news.rcn.net!rcn!207-172-255-2 Xref: chonsp.franklin.ch alt.folklore.computers:53474 In article , Mike Swaim wrote: >In alt.folklore.computers jmfbahciv@aol.com wrote: >> Reading any file that is non-local dictates that the file >> gets shipped to the local host...always. > > Where on the local host? What if it doesn't have local disk, and the >file's larger than RAM? > It does if the compile command indicates that the compilation is supposed to happen on the local host (that's what I thought Peter wanted). If you don't have the capacity for the compilation, then I guess you're not asking to have the work done on the local host. There should be (at least) two options for the user: compile the specified program on the remote host or compile the program on the local host. And you can get as complicated as you want _as long as a full file specification is supported in the user network interface. The whole point I'm trying to make is that the user be _allowed_ to have the option. /BAH Subtract a hundred and four for e-mail. ###### From: jmfbahciv@aol.com Newsgroups: alt.sys.pdp8,alt.sys.pdp10,alt.folklore.computers Subject: Re: Filenames, shells, remote file access [was: ooh, a *real* flamewar :) Date: Sat, 01 Apr 00 09:38:40 GMT Organization: UltraNet Communications, Inc. Lines: 38 Message-ID: <8c4lr9$ksm$2@bob.news.rcn.net> References: <8bvbie$20t$1@bob.news.rcn.net> <8c20an$npb$1@bob.news.rcn.net> <8c2p29$8pp$1@hawkins.cba.uni.edu> X-Trace: rVpZd0i/SFC8nmBwAsPfb3yCBbqpj4Ll0Y8q0i3zQro= X-Complaints-To: abuse@rcn.com NNTP-Posting-Date: 1 Apr 2000 11:16:57 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.maxwell.syr.edu!feed1.news.rcn.net!rcn!207-172-255-109 Xref: chonsp.franklin.ch alt.folklore.computers:53479 In article <8c2p29$8pp$1@hawkins.cba.uni.edu>, hawk@hawkins.cba.uni.edu wrote: >In article <8c20an$npb$1@bob.news.rcn.net>, wrote: > >>It does if the compile command indicates that the compilation is >>supposed to happen on the local host (that's what I thought >>Peter wanted). If you don't have the capacity for the compilation, >>then I guess you're not asking to have the work done on the local >>host. > >Yikes, toss this machine off my desk . . . :) [hmm, not a bad >idea anyway. . . .] > >I'm using Debian Linux on this box instead of the FreeBSD I prefer >precisely because I don't have the resources to compile things, >and need everything precompiled. The package manager is too much >a load on this thing . . . (Yes, I know about the packages in bsd. >Not only do I normally use them, but I compile them. However, I >need my customized version of LyX, and am forced to do a linux >compile, and the other machines I can reacxh during the day to >do this are Linux, not bsd.) Thus, in your case, due to hardware constraints, you need the compilation to happen over there somewhere. However, you also know where that compilation has to happen in order to generate code that can be run on your machine. Now, the command to specify which flavor of CPU you're building the EXE for can be a full file specification or a switch. I don't really care about syntax of the intent but it is still the _user_ who is stating how the thing is getting built. And that's all I'm talking about. Nothing happens automagically...it just seems that way. I'm discovering that this Misoft mange is more invasive than I thought. /BAH Subtract a hundred and four for e-mail. ###### From: jmfbahciv@aol.com Newsgroups: alt.sys.pdp8,alt.sys.pdp10,alt.folklore.computers Subject: Re: Filenames, shells, remote file access [was: ooh, a *real* flamewar :) Date: Sat, 01 Apr 00 09:34:34 GMT Organization: UltraNet Communications, Inc. Lines: 26 Message-ID: <8c4ljj$ksm$1@bob.news.rcn.net> References: <8bvbie$20t$1@bob.news.rcn.net> <8c20an$npb$1@bob.news.rcn.net> <8c2ih5$2m34$1@citadel.in.taronga.com> X-Trace: TxU8RNi6Q4laPhzUYBdmUMg/XtsFSfOWqeOavHPrLpg= X-Complaints-To: abuse@rcn.com NNTP-Posting-Date: 1 Apr 2000 11:12:51 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!207-172-255-109 Xref: chonsp.franklin.ch alt.folklore.computers:53482 In article <8c2ih5$2m34$1@citadel.in.taronga.com>, peter@taronga.com (Peter da Silva) wrote: >In article <8c20an$npb$1@bob.news.rcn.net>, wrote: >>It does if the compile command indicates that the compilation is >>supposed to happen on the local host (that's what I thought >>Peter wanted). If you don't have the capacity for the compilation, >>then I guess you're not asking to have the work done on the local >>host. > >You don't have to have enough local storage to hold the files you're >compiling. At one time it was very popular to have workstations with *no* >local disk storage... all files, programs, even swap, were accessed over >the network. Then disks got really cheap and that trade-off no longer made >sense... but it would have been really really annoying to have to stick >an expensive hard disk inside every box just so it could hold temporary >copies of files. > You still don't get it. It all depends on where you want the compilation to take place. A compilation on a PDP-11 will generate much different code than a compilation on a 486. The computer has no idea where the executable is going to be run, now does it? /BAH Subtract a hundred and four for e-mail. ###### From: "Simo Tuominen" Newsgroups: alt.sys.pdp8,alt.sys.pdp10,alt.folklore.computers Subject: Re: Filenames, shells, remote file access [was: ooh, a *real* flamewar :) Date: Thu, 30 Mar 2000 18:50:30 +0300 Organization: Freenet Finland Lines: 41 Message-ID: <8c1kma$1enq$1@learnet.freenet.hut.fi> References: <8bqmgn$1pas$1@citadel.in.taronga.com> <8bsv9i$9kv$2@bob.news.rcn.net> <8bthc5$2d2i$1@citadel.in.taronga.com> <8bvbie$20t$1@bob.news.rcn.net> NNTP-Posting-Host: aqua00.edu2.evitech.fi 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 Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!newsfeeds.belnet.be!news.belnet.be!skynet.be!news.algonet.se!algonet!news-peer-europe.sprintlink.net!news.sprintlink.net!newsfeed.bahnhof.se!news.defero.net!news.bbnetworks.net!learnet.freenet.hut.fi!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:53538 wrote in message news:8bvbie$20t$1@bob.news.rcn.net... > In article <8bthc5$2d2i$1@citadel.in.taronga.com>, > peter@taronga.com (Peter da Silva) wrote: > >Um, that's not what I think of as network file access. Why > >should it have to ship a file over here before I even start > >to read it? Hell, I may not have even opened it yet... > > Reading any file that is non-local dictates that the file > gets shipped to the local host...always. (I have to say that some of your views are ...surprising.) It may also mean that only the bytes being read are transfered, which, granted, isn't always the case. Not all systems/programs can handle streaming media. > >Or do you mean that every time I run the COBOL compiler I have to tell it > >the destination file name as well as the source file name? > > Of course you do. Just because you don't explicitly give > a file destination doesn't mean that you haven't implicitly > specified it...we used call that filling in the defaults. That's silly. IMO if you compile something from source files that are on a network volume the compiler shall (*not* "should") access the files where they are and put the resulting files where I tell it to. There are *no* local names for any of the source files. > >I would personally consider that user-hostile. > > Computers don't read minds but they do make assumptions when > necessary data isn't specified by the users. User hostility > is when that data is filled in without giving the user the > option. Stupid behavior is also user-hostile, whatever the reason. ###### From: "Simo Tuominen" Newsgroups: alt.sys.pdp8,alt.sys.pdp10,alt.folklore.computers Subject: Re: Filenames, shells, remote file access [was: ooh, a *real* flamewar :) Date: Thu, 30 Mar 2000 18:50:30 +0300 Organization: Freenet Finland Lines: 41 Message-ID: <8c1kkk$fqk6$1@learnet.freenet.hut.fi> References: <8bqmgn$1pas$1@citadel.in.taronga.com> <8bsv9i$9kv$2@bob.news.rcn.net> <8bthc5$2d2i$1@citadel.in.taronga.com> <8bvbie$20t$1@bob.news.rcn.net> NNTP-Posting-Host: aqua00.edu2.evitech.fi 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 Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news.maxwell.syr.edu!news.algonet.se!newsfeed1.telenordia.se!algonet!newsfeed.bahnhof.se!news.defero.net!news.bbnetworks.net!learnet.freenet.hut.fi!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:53542 wrote in message news:8bvbie$20t$1@bob.news.rcn.net... > In article <8bthc5$2d2i$1@citadel.in.taronga.com>, > peter@taronga.com (Peter da Silva) wrote: > >Um, that's not what I think of as network file access. Why > >should it have to ship a file over here before I even start > >to read it? Hell, I may not have even opened it yet... > > Reading any file that is non-local dictates that the file > gets shipped to the local host...always. (I have to say that some of your views are ...surprising.) It may also mean that only the bytes being read are transfered, which, granted, isn't always the case. Not all systems/programs can handle streaming media. > >Or do you mean that every time I run the COBOL compiler I have to tell it > >the destination file name as well as the source file name? > > Of course you do. Just because you don't explicitly give > a file destination doesn't mean that you haven't implicitly > specified it...we used call that filling in the defaults. That's silly. IMO if you compile something from source files that are on a network volume the compiler shall (*not* "should") access the files where they are and put the resulting files where I tell it to. There are *no* local names for any of the source files. > >I would personally consider that user-hostile. > > Computers don't read minds but they do make assumptions when > necessary data isn't specified by the users. User hostility > is when that data is filled in without giving the user the > option. Stupid behavior is also user-hostile, whatever the reason. ###### From: Mike Swaim Subject: Re: Filenames, shells, remote file access [was: ooh, a *real* flamewar :) Newsgroups: alt.sys.pdp8,alt.sys.pdp10,alt.folklore.computers References: <8bqmgn$1pas$1@citadel.in.taronga.com> <8bsv9i$9kv$2@bob.news.rcn.net> <8bthc5$2d2i$1@citadel.in.taronga.com> <8bvbie$20t$1@bob.news.rcn.net> User-Agent: tin/pre-1.4-19990413 ("Endemoniada") (UNIX) (FreeBSD/3.2-RELEASE (i386)) Lines: 10 Message-ID: NNTP-Posting-Host: 207.91.187.171 X-Trace: tw11.nn.bcandid.com 954491245 207.91.187.171 (Fri, 31 Mar 2000 01:27:25 MST) NNTP-Posting-Date: Fri, 31 Mar 2000 01:27:25 MST Organization: bCandid - Powering the world's discussions - http://bCandid.com Date: Fri, 31 Mar 2000 08:27:25 GMT Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!isdnet!howland.erols.net!portc.blue.aol.com.MISMATCH!portc01.blue.aol.com!portc.blue.aol.com!gw12.bcandid.com.MISMATCH!gw22.nn.bcandid.com!hub12.nn.bcandid.com!tw11.nn.bcandid.com.POSTED!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:53526 In alt.folklore.computers jmfbahciv@aol.com wrote: > Reading any file that is non-local dictates that the file > gets shipped to the local host...always. Where on the local host? What if it doesn't have local disk, and the file's larger than RAM? -- Mike Swaim, Avatar of Chaos: Disclaimer:I sometimes lie. Home: swaim at nol * net Quote: "Boingie"^4 Y,W&D ###### From: peter@taronga.com (Peter da Silva) Newsgroups: alt.sys.pdp8,alt.sys.pdp10,alt.folklore.computers Subject: Re: Filenames, shells, remote file access [was: ooh, a *real* flamewar :) Date: 31 Mar 2000 16:08:05 GMT Organization: TSS Inc. Lines: 20 Message-ID: <8c2ih5$2m34$1@citadel.in.taronga.com> References: <8bvbie$20t$1@bob.news.rcn.net> <8c20an$npb$1@bob.news.rcn.net> NNTP-Posting-Host: citadel.in.taronga.com X-Trace: citadel.in.taronga.com 954518885 88164 10.0.0.43 (31 Mar 2000 16:08:05 GMT) X-Complaints-To: usenet@taronga.com NNTP-Posting-Date: 31 Mar 2000 16:08:05 GMT X-Newsreader: trn 4.0-test72 (19 April 1999) Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!newsfeed-zh.ip-plus.net!news.ip-plus.net!nyc-news-feed1.bbnplanet.com!nycmny1-snh1.gtei.net!paloalto-snf1.gtei.net!news.gtei.net!newsfeed.stanford.edu!news-proxy.baileynm.com!citadel.in.taronga.com!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:53521 In article <8c20an$npb$1@bob.news.rcn.net>, wrote: >It does if the compile command indicates that the compilation is >supposed to happen on the local host (that's what I thought >Peter wanted). If you don't have the capacity for the compilation, >then I guess you're not asking to have the work done on the local >host. You don't have to have enough local storage to hold the files you're compiling. At one time it was very popular to have workstations with *no* local disk storage... all files, programs, even swap, were accessed over the network. Then disks got really cheap and that trade-off no longer made sense... but it would have been really really annoying to have to stick an expensive hard disk inside every box just so it could hold temporary copies of files. -- This is The Reverend Peter da Silva's Boring Sig File - there are no references to Wolves, Kibo, Discordianism, or The Church of the Subgenius in this document Executive Vice President, Corporate Communications, Entropy Gradient Reversals. ###### From: hawk@hawkins.cba.uni.edu Newsgroups: alt.sys.pdp8,alt.sys.pdp10,alt.folklore.computers Subject: Re: Filenames, shells, remote file access [was: ooh, a *real* flamewar :) Date: 31 Mar 2000 11:59:37 -0600 Organization: House of Hawkins Lines: 27 Message-ID: <8c2p29$8pp$1@hawkins.cba.uni.edu> References: <8bvbie$20t$1@bob.news.rcn.net> <8c20an$npb$1@bob.news.rcn.net> NNTP-Posting-Host: hawkins.cba.uni.edu Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!newscore.univie.ac.at!newsfeed.Austria.EU.net!nmaster.kpnqwest.net!npeer.kpnqwest.net!cpk-news-hub1.bbnplanet.com!news.gtei.net!news.idt.net!logbridge.uoregon.edu!news.physics.uiowa.edu!news.uni.edu!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:53498 In article <8c20an$npb$1@bob.news.rcn.net>, wrote: >It does if the compile command indicates that the compilation is >supposed to happen on the local host (that's what I thought >Peter wanted). If you don't have the capacity for the compilation, >then I guess you're not asking to have the work done on the local >host. Yikes, toss this machine off my desk . . . :) [hmm, not a bad idea anyway. . . .] I'm using Debian Linux on this box instead of the FreeBSD I prefer precisely because I don't have the resources to compile things, and need everything precompiled. The package manager is too much a load on this thing . . . (Yes, I know about the packages in bsd. Not only do I normally use them, but I compile them. However, I need my customized version of LyX, and am forced to do a linux compile, and the other machines I can reacxh during the day to do this are Linux, not bsd.) hawk -- Prof. Richard E. Hawkins, Esq. hawk@hawkins.cba.uni.edu (319) 266-7114 http://eyry.econ.iastate.edu/hawk These opinions will not be those of UNI until it pays my retainer. ###### From: michael.wojcik@merant.com (Michael Wojcik) Newsgroups: alt.sys.pdp8,alt.sys.pdp10,alt.folklore.computers Subject: Re: Filenames, shells, remote file access [was: ooh, a *real* flamewar :) Followup-To: alt.folklore.computers Date: 31 Mar 2000 19:31:52 GMT Organization: MERANT Inc. Lines: 61 Message-ID: <8c2uf801tas@news2.newsguy.com> References: <8bsv9i$9kv$2@bob.news.rcn.net> <8bthc5$2d2i$1@citadel.in.taronga.com> <8bvbie$20t$1@bob.news.rcn.net> <8bvoo7$r6i$1@citadel.in.taronga.com> Reply-To: michael.wojcik@merant.com NNTP-Posting-Host: p-190.newsdawg.com X-Newsreader: xrn 9.00 Originator: mww@lorelei-n Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!enews.sgi.com!pln-w!spln!extra.newsguy.com!newsp.newsguy.com!mww Xref: chonsp.franklin.ch alt.folklore.computers:53495 [Followups de-PDP'd.] In article <8bvoo7$r6i$1@citadel.in.taronga.com>, peter@taronga.com (Peter da Silva) writes: > In article <8bvbie$20t$1@bob.news.rcn.net>, wrote: > >Reading any file that is non-local dictates that the file > >gets shipped to the local host...always. > Why? > > That's what the VICE file system in Andrew did, and I believe AFS still does > this, and I suppose for most cases that's good enough. It really breaks things > when you're doing something like looking at the last few lines of an 11 meg > logfile or grabbing a couple of records from a database. Are you sure about this? I used VICE/AFS for years, and I don't recall it necessarily caching the entire file. It's been a while, though. In any case, the many advantages of AFS over NFS (like supporting transparent backup servers, Kerberos integration, ACLs, keeping state rather than pretending not keeping it is a good idea, and, um, local file caching) outweighed the storage cost of the local cache, even with 1989 disk sizes, and our LAN didn't see excessive traffic because of it. Personally, I wouldn't use a network file system for direct file- level access to a database - that's what networked DBMSes and distributed applications are for. I want to restrict access to the store itself. > And in any case this should be handled behind the program's back. To the user > it looks like he's directly accessing the file on the remote box. Since both NFS and AFS are mountable under Unix, to the user it looks like directly accessing the file on the local box, really. And that, of course, provides a single syntax for file access, which is what this discussion is all about. In general, I think that's a Good Thing, but there are times when you should have access to the remote system's native naming convention. For example: I end up shipping files back and forth between the AS/400 and operating systems not invented by incomprehensible space aliens now and then, and I frequently use FTP rather than networked file systems to do so, largely because FTP lets me specify destination syntax in OS/400 IFS form. -- Michael Wojcik michael.wojcik@merant.com AAI Development, MERANT (block capitals are a company mandate) Department of English, Miami University However, we maintain that our mission is more than creating high-tech amusement--rather, we must endeavor to provide high-tech, high-touch entertainment with an emphasis on enkindling human warmth. -- "The Ultimate in Entertainment", from the president of video game producer Namco ###### From: peter@taronga.com (Peter da Silva) Newsgroups: alt.sys.pdp8,alt.sys.pdp10,alt.folklore.computers Subject: Re: Filenames, shells, remote file access [was: ooh, a *real* flamewar :) Date: 1 Apr 2000 13:20:17 GMT Organization: TSS Inc. Lines: 55 Message-ID: <8c4t2h$196g$1@citadel.in.taronga.com> References: <8c20an$npb$1@bob.news.rcn.net> <8c2ih5$2m34$1@citadel.in.taronga.com> <8c4ljj$ksm$1@bob.news.rcn.net> NNTP-Posting-Host: citadel.in.taronga.com X-Trace: citadel.in.taronga.com 954595217 42192 10.0.0.43 (1 Apr 2000 13:20:17 GMT) X-Complaints-To: usenet@taronga.com NNTP-Posting-Date: 1 Apr 2000 13:20:17 GMT X-Newsreader: trn 4.0-test72 (19 April 1999) Path: chonsp.franklin.ch!pfaff.ethz.ch!news-zh.switch.ch!news-ge.switch.ch!news.maxwell.syr.edu!newsfeed.stanford.edu!news-proxy.baileynm.com!citadel.in.taronga.com!not-for-mail Xref: chonsp.franklin.ch alt.folklore.computers:53578 In article <8c4ljj$ksm$1@bob.news.rcn.net>, wrote: >In article <8c2ih5$2m34$1@citadel.in.taronga.com>, > peter@taronga.com (Peter da Silva) wrote: >>In article <8c20an$npb$1@bob.news.rcn.net>, wrote: >>>It does if the compile command indicates that the compilation is >>>supposed to happen on the local host (that's what I thought >>>Peter wanted). If you don't have the capacity for the compilation, >>>then I guess you're not asking to have the work done on the local >>>host. >>You don't have to have enough local storage to hold the files you're >>compiling. At one time it was very popular to have workstations with *no* >>local disk storage... all files, programs, even swap, were accessed over >>the network. Then disks got really cheap and that trade-off no longer made >>sense... but it would have been really really annoying to have to stick >>an expensive hard disk inside every box just so it could hold temporary >>copies of files. >You still don't get it. Sure I do. >It all depends on where you want >the compilation to take place. I want the compilation to take place on my workstation, even though the files are stored on the VAX over there because theat's where I have space for them. >A compilation on a PDP-11 >will generate much different code than a compilation on >a 486. The computer has no idea where the executable is going >to be run, now does it? I'm not sure I see the relevance. I used a COBOL compiler as an example, but it could equally well be a statistics package like SAS or SPSS, or a circuit simulator like SPICE, or a text processor like TROFF or RUNOFF, or some crufty crystallography or finite element analysis code. The point I'm making is that programs need to be able to parse file names and generate output files or other related files from the names of the input files, and it shouldn't matter whether the files are local or remote. File name syntax really should be seen as part of the API. People do this all the time. Hell, we had code running on RMX-86 (with a file name syntax of :dev:file.ext) saving files to an assigned name that translated to a named pipe on a Xenix-286 box that was being read by the program that did the next stage of the processing. To the RMX-86 program it was just :log:pipe.1 or whatever, and on Xenix it was //xdsXX/home/logger/pipe.1. -- This is The Reverend Peter da Silva's Boring Sig File - there are no references to Wolves, Kibo, Discordianism, or The Church of the Subgenius in this document Executive Vice President, Corporate Communications, Entropy Gradient Reversals.