From: Jesus Monroy Jr (jmonroy@netcom.com) Subject: We plead again. Newsgroups: comp.os.386bsd.development View: (This is the only article in this thread) | Original Format Date: 1993-04-24 01:21:27 PST Responses to "Plea across the board" =================================== Again... OK..... I will state this in a technical manner so you may understand me. The problem arises in the releases of 386BSD, not just the one's done by Bill & Lynn, but by others (NETBSD, patchkits, etc.). The problem is some people do not have room for all the binaries on their machines. They use it in place of the lost VAX 11/750 or maybe as an learning tool. They, and others, become confused when conflicting version of an applicaiton become available. Why is this important? Since they do not have the source or a source tree, they are unable to make repairs as we do. Hence, solutions like rcs & sccs are not useful. A good solution might be that a version number could be available from a program at run-time. Again, version numbers in a binary format would provide, if not a good solution, a reasonal universal pointer. I am not trying to tell you what to do. I am only trying to make you see the raw nature of the problem. ========================================================================= >> From: overby@cray.com (Glen Overby) >> Organization: Lego Land >> >> In article jmonroy@netcom.com (Jesus Monroy Jr) writes: >> > A Plea follows to 386BSD developers >> > After several frustraiong experiences with linux software, I'd >> >like to propose that all linux authors and porters try to put a version >> >number in their binaries that can be called with -v, -V, -h or -?. It >> >> I'd like to suggest something similar, yet different: >> >> (1) use a source control mechanism such as RCS/CVS or SCCS >> NO, This I beleive is a useless solution, especially for binary run-time problems. >> (2) define a "static" character array in each source file that >> contains the source control system's version identification. This >> allows anyone with read-access to the executable to find the version >> numbers of _all_ source modules making up the executable using >> Again, some people do not have the source. ========================================================================= >> From: cgd@gaia.CS.Berkeley.EDU (Chris G. Demetriou) >> Organization: Kernel Hackers 'r' Us >> >> In article <1993Apr20.102524.4417@hemlock.cray.com> overby@cray.com (Glen Overby) writes: >> >(2) define a "static" character array in each source file that >> >contains the source control system's version identification. This >> >allows anyone with read-access to the executable to find the version >> >numbers of _all_ source modules making up the executable using >> >> we're going to be using rcsid strings for NetBSD... >> I would ask that you add to that with, maybe something that is available in a run-time situation. ========================================================================= >> From: peter@NeoSoft.com (Peter da Silva) >> Date: Wed, 21 Apr 1993 01:51:27 GMT >> >> In article jmonroy@netcom.com (Jesus Monroy Jr) >> passes on a plea: >> > After several frustraiong experiences with linux software, I'd >> > like to propose that all linux authors and porters try to put a version >> > number in their binaries that can be called with -v, -V, -h or -?. >> >> Put a string in that can be grepped for with "what". Right now, the output >> This is not a solution either. Adding whatever binary garbage is net to the string would only add confusion, But it may point to a similarly good idea. I will let you hash it out. ========================================================================= >> From: terry@cs.weber.edu (A Wizard of Earth C) >> Date: Wed, 21 Apr 93 20:05:07 GMT >> >> To answer the question instead of the plea: >> >> The only viable option (not taken already by some command, thus causing an >> incompatability with Posix, existing shell scripts, etc.) is the "-?". This >> has the unfortunate characteristic of being a globbing character in many >> popular shells (which probably explains why it isn't taken). >> Terry, Thank you for helping make some sense of this issue. I was aware of this, but I think others are not. >> Take a look at 1003.2. I think the loss of standards compliance and of >> backward compatability with existing code is too high a price to pay for >> people who probably are going to guess "/h" or "/help" anyway. >> Wow, I love talk like this. >> To assume they know the UNIX options delimiters is as large an >> assumption as expecting them to know about "man". >> More enlighting stuff. >> If typing "man" is "too cryptic", ln it to "help" on your system; if using >> "man" is too cryptic, archie for "help"... there are many excellent packages. >> To exaggerate a [very] little, every third Senior project is a help system. >> I am confused. What does this mean: "every third Senior project is a help system" ? I did not mean to say that everyone should rewrite "man" or start some other system. ========================================================================= >> From: storm@cs.mcgill.ca (Marc WANDSCHNEIDER) >> Date: Thu, 22 Apr 1993 00:14:12 GMT >> >> In article <1993Apr21.200507.7489@fcom.cc.utah.edu> terry@cs.weber.edu (A Wizard of Earth C) writes: >> >If typing "man" is "too cryptic", ln it to "help" on your system; if using >> >"man" is too cryptic, archie for "help"... there are many excellent packages. >> >To exaggerate a [very] little, every third Senior project is a help system. >> >> It also might be an idea to support the way GNU is doing it with their >> programs these days (which I believe is POSIX motivated)---with the >> >> --version flag. >> WOW, at least a hope of decency! ========================================================================= ___________________________________________________________________________ Jesus Monroy Jr jmonroy@netcom.com /386BSD/device-drivers /fd /qic /clock /documentation ___________________________________________________________________________ From: Jesus Monroy Jr (jmonroy@netcom.com) Subject: PLEA across the board. Newsgroups: comp.os.386bsd.development View this article only Date: 1993-04-20 02:45:41 PST A Plea follows to 386BSD developers ----------------------------------- The following two message came from the LINUX group. I am making the same call to 386BSD developers. ========================================================================== From: jhenders@wimsey.bc.ca (John Henders) Subject: Plea to linux developers! Organization: Wimsey Information Services Date: Sat, 17 Apr 1993 20:07:10 GMT After several frustraiong experiences with linux software, I'd like to propose that all linux authors and porters try to put a version number in their binaries that can be called with -v, -V, -h or -?. It would be to everyone's benifit if there was an easy way to find out what version of a package, instead of sometimes having to go by the file date or what directory on what archive site it was on. What do people think? -- John Henders ========================================================================== From: jeffk@cyberspace.org (Jeff Kopmanis) Subject: Re: Plea to linux developers! Organization: GREX Public Access Unix +1 313 761 3000 Date: Mon, 19 Apr 1993 01:59:01 GMT I agree wholeheartedly! I'd even go so far as to ask developers to put in a short description of the program and what its supposed to do. Granted, most man pages have this sort of thing, but often stuff doesn't come with man pages. If developers don't think this is a great idea, then at least submit something to the LSM (Linux Software Map). That way the info is at least documented somewhere. BTW: I'm still taking volunteers for helping with the LSM. Mail me if you're interested in lending a hand! (jeffk@cyberspace.org) -- --------------------------------------------------------------------------- Jeff Kopmanis jeffk@cyberspace.org (313) 769-6911 (home) (313) 393-4784 (work) ___________________________________________________________________________ Jesus Monroy Jr jmonroy@netcom.com /386BSD/device-drivers /fd /qic /clock /documentation ___________________________________________________________________________ Message 2 in thread From: Glen Overby (overby@cray.com) Subject: Re: PLEA across the board. Newsgroups: comp.os.386bsd.development View this article only Date: 1993-04-20 09:28:11 PST In article jmonroy@netcom.com (Jesus Monroy Jr) writes: > A Plea follows to 386BSD developers >After several frustraiong experiences with linux software, I'd >like to propose that all linux authors and porters try to put a version >number in their binaries that can be called with -v, -V, -h or -?. It I'd like to suggest something similar, yet different: (1) use a source control mechanism such as RCS/CVS or SCCS (2) define a "static" character array in each source file that contains the source control system's version identification. This allows anyone with read-access to the executable to find the version numbers of _all_ source modules making up the executable using rcsident(1) sccswhat(1) Ditto with libraries. Header files are tougher because of multiple references. Then you don't have to worry about assigning and updating an over-all program version. Just make sure you distribute binaries build from sources that haven't been modified since their most recent check-out. Glen/B Message 3 in thread From: Chris G. Demetriou (cgd@gaia.CS.Berkeley.EDU) Subject: Re: PLEA across the board. Newsgroups: comp.os.386bsd.development View this article only Date: 1993-04-20 11:51:58 PST In article <1993Apr20.102524.4417@hemlock.cray.com> overby@cray.com (Glen Overby) writes: >(2) define a "static" character array in each source file that >contains the source control system's version identification. This >allows anyone with read-access to the executable to find the version >numbers of _all_ source modules making up the executable using we're going to be using rcsid strings for NetBSD... the only problem is, if you "cvs export" a tree, the dollar signs get stripped off the rcs strings, (for good reason), and that's what ident looks for... chris -- Chris G. Demetriou cgd@cs.berkeley.edu "386bsd as depth first search: whenever you go to fix something you find that 3 more things are actually broken." -- Adam Glass Message 4 in thread From: Michael Meissner (meissner@osf.org) Subject: Re: PLEA across the board. Newsgroups: comp.os.386bsd.development View this article only Date: 1993-04-23 18:55:45 PST In article <1993Apr20.102524.4417@hemlock.cray.com> overby@cray.com (Glen Overby) writes: | I'd like to suggest something similar, yet different: | | (1) use a source control mechanism such as RCS/CVS or SCCS | | (2) define a "static" character array in each source file that | contains the source control system's version identification. This | allows anyone with read-access to the executable to find the version | numbers of _all_ source modules making up the executable using | | rcsident(1) | sccswhat(1) There is a general problem with this scheme (at least for RCS) unless you are careful to use either non-standard keywords or something like co -kv to make releases. The problem is, if the user then puts the software under the same source control system, the keywords will be updated to be the local version number (ie, 1.1), rather than the release version number. This makes the situation even worse if the application did use the rcs keywords to generate version numbers or ident/what strings. -- Michael Meissneremail: meissner@osf.orgphone: 617-621-8861 Open Software Foundation, 11 Cambridge Center, Cambridge, MA, 02142 You are in a twisty little passage of standards, all conflicting. Message 5 in thread From: Peter da Silva (peter@NeoSoft.com) Subject: Re: PLEA across the board. Newsgroups: comp.os.386bsd.development, comp.os.linux View this article only Date: 1993-04-20 22:23:26 PST In article jmonroy@netcom.com (Jesus Monroy Jr) passes on a plea: > After several frustraiong experiences with linux software, I'd > like to propose that all linux authors and porters try to put a version > number in their binaries that can be called with -v, -V, -h or -?. Put a string in that can be grepped for with "what". Right now, the output from "what" is a little verbose (it gives you the version number of every module in the file), but a "what -q" that just gives strings flagged with a "program version number" would be a trivial enhancement, then you could do "what /bin/*" and see what version of *everything* you're using. Much better than adding an option (that might conflict with existing options) to every program. -- Peter da Silva. . `-_-' Oletko halannut suttasi tänään? 'U` Tarjoilija, tämä ateria elää vielä. Message 6 in thread From: Stephen Tweedie (sct@dcs.ed.ac.uk) Subject: Re: PLEA across the board. Newsgroups: comp.os.386bsd.development, comp.os.linux View this article only Date: 1993-04-21 23:07:17 PST In article , peter@NeoSoft.com (Peter da Silva) writes: > Put a string in that can be grepped for with "what". Right now, the > output from "what" is a little verbose (it gives you the version > number of every module in the file), but a "what -q" that just gives > strings flagged with a "program version number" would be a trivial > enhancement, then you could do "what /bin/*" and see what version of > *everything* you're using. > Much better than adding an option (that might conflict with existing > options) to every program. Why not just do it the RCS way, and have a "$Id:$" string in any modules you want to identify. This way, the information is automatically kept uptodate by RCS or CVS, and you can use the existing "ident" command from RCS to identify both source and code (assuming you have $Id$ string constants in your modules). Cheers, Stephen Tweedie. --- Stephen Tweedie (Internet: ) Department of Computer Science, Edinburgh University, Scotland. Message 7 in thread From: David W. Summers (dws@cseg03.engr.uark.edu) Subject: Re: PLEA across the board. Newsgroups: comp.os.386bsd.development, comp.os.linux View this article only Date: 1993-04-22 20:34:48 PST To: sct@dcs.ed.ac.uk Subject: Re: PLEA across the board. Newsgroups: comp.os.386bsd.development,comp.os.linux References: In comp.os.linux you write: >In article , peter@NeoSoft.com (Peter da Silva) writes: > >> Put a string in that can be grepped for with "what". Right now, the >> output from "what" is a little verbose (it gives you the version >> number of every module in the file), but a "what -q" that just gives >> strings flagged with a "program version number" would be a trivial >> enhancement, then you could do "what /bin/*" and see what version of >> *everything* you're using. >> Much better than adding an option (that might conflict with existing >> options) to every program. > >Why not just do it the RCS way, and have a "$Id:$" string in any >modules you want to identify. This way, the information is >automatically kept uptodate by RCS or CVS, and you can use the >existing "ident" command from RCS to identify both source and code >(assuming you have $Id$ string constants in your modules). > >Cheers, > Stephen Tweedie. >--- >Stephen Tweedie (Internet: ) >Department of Computer Science, Edinburgh University, Scotland. I have a short program that will search for both "what" (SCCS) strings and "ident" (RCS) strings at the same time. I wrote it a few years ago as I began switching over to RCS from SCCS. I can e-mail or post it if anyone is interested. I find it quite handy. - David Summers (dws@engr.uark.edu) -- "Never under-estimate the bandwidth of a station-wagon David Summers full of tapes, hurtling down the highway." dws@engr.uark.edu - Tanenbaum, "Computer Networks" Message 8 in thread From: Peter da Silva (peter@NeoSoft.com) Subject: Re: PLEA across the board. Newsgroups: comp.os.386bsd.development, comp.os.linux View this article only Date: 1993-04-23 18:34:10 PST In article sct@dcs.ed.ac.uk (Stephen Tweedie) writes: > In article , peter@NeoSoft.com (Peter da Silva) writes: > > Put a string in that can be grepped for with "what". > Why not just do it the RCS way, and have a "$Id:$" string in any > modules you want to identify. That's exactly the *same* solution, except for RCS instead of SCCS. The problem with "what" and "ident" is they give you *all* the ID strings in the program, rather than just the one for the mainline. Hence my suggestion for a "what -q" (or ident -q). -- Peter da Silva. . `-_-' Har du kramat din varg idag? 'U` My Apple-II has more RAM than my Mac! Message 9 in thread From: Chris G. Demetriou (cgd@gaia.CS.Berkeley.EDU) Subject: Re: PLEA across the board. Newsgroups: comp.os.386bsd.development, comp.os.linux View this article only Date: 1993-05-06 05:36:40 PST In article sct@dcs.ed.ac.uk (Stephen Tweedie) writes: =>Why not just do it the RCS way, and have a "$Id:$" string in any =>modules you want to identify. This way, the information is =>automatically kept uptodate by RCS or CVS, and you can use the =>existing "ident" command from RCS to identify both source and code =>(assuming you have $Id$ string constants in your modules). because when you "cvs export" a source tree, by default, it gets rid of the dollar-signs, to make the version numbers of an exported release "static" and importable into another person's/group's environment... i, for one, think this is a *very* good+important thing, but it breaks ident... chris -- Chris G. Demetriou cgd@cs.berkeley.edu "386bsd as depth first search: whenever you go to fix something you find that 3 more things are actually broken." -- Adam Glass Message 10 in thread From: A Wizard of Earth C (terry@cs.weber.edu) Subject: Re: PLEA across the board. Newsgroups: comp.os.386bsd.development View this article only Date: 1993-04-21 14:31:59 PST In article jmonroy@netcom.com (Jesus Monroy Jr) writes: > > A Plea follows to 386BSD developers > ----------------------------------- > > The following two message came from the LINUX group. > > I am making the same call to 386BSD developers. > > >========================================================================== >From: jhenders@wimsey.bc.ca (John Henders) >Subject: Plea to linux developers! >Organization: Wimsey Information Services >Date: Sat, 17 Apr 1993 20:07:10 GMT > > >After several frustraiong experiences with linux software, I'd >like to propose that all linux authors and porters try to put a version >number in their binaries that can be called with -v, -V, -h or -?. It >would be to everyone's benifit if there was an easy way to find out what >version of a package, instead of sometimes having to go by the file date >or what directory on what archive site it was on. >What do people think? To answer the question instead of the plea: The only viable option (not taken already by some command, thus causing an incompatability with Posix, existing shell scripts, etc.) is the "-?". This has the unfortunate characteristic of being a globbing character in many popular shells (which probably explains why it isn't taken). Take a look at 1003.2. I think the loss of standards compliance and of backward compatability with existing code is too high a price to pay for people who probably are going to guess "/h" or "/help" anyway. To assume they know the UNIX options delimiters is as large an assumption asexpecting them to know about "man". If typing "man" is "too cryptic", ln it to "help" on your system; if using "man" is too cryptic, archie for "help"... there are many excellent packages. To exaggerate a [very] little, every third Senior project is a help system. Terry Lambert terry@icarus.weber.edu --- Any opinions in this posting are my own and not those of my present or previous employers. -- ------------------------------------------------------------------------------- "I have an 8 user poetic license" - me Get the 386bsd FAQ from agate.berkeley.edu:/pub/386BSD/386bsd-0.1/unofficial ------------------------------------------------------------------------------- From: Dave Burgess (burgess@hrd769.brooks.af.mil) Subject: Re: PLEA across the board. Newsgroups: comp.os.386bsd.development View this article only Date: 1993-04-21 15:02:47 PST In article <1993Apr21.200507.7489@fcom.cc.utah.edu> terry@cs.weber.edu (A Wizard of Earth C) writes: > >If typing "man" is "too cryptic", ln it to "help" on your system; if using >"man" is too cryptic, archie for "help"... there are many excellent packages. >To exaggerate a [very] little, every third Senior project is a help system. > > I found that a much more 'help'ful link was to 'ln' apropos to help. This was good for two reasons. The first (and most loudly exclaimed) was that no one could remember how to spell apropos. The second was that I needed a bargaining chit to get folks to start learning about Unix and stop whining because it wasn't a VAX running VMS... -- ------ TSgt Dave Burgess NCOIC AL/Management Information Systems Office Brooks AFB, TX Message 12 in thread From: Marc WANDSCHNEIDER (storm@cs.mcgill.ca) Subject: Re: PLEA across the board. Newsgroups: comp.os.386bsd.development View this article only Date: 1993-04-21 17:42:36 PST In article <1993Apr21.200507.7489@fcom.cc.utah.edu> terry@cs.weber.edu (A Wizard of Earth C) writes: >In article jmonroy@netcom.com (Jesus Monroy Jr) writes: >> >> A Plea follows to 386BSD developers >> ----------------------------------- >>like to propose that all linux authors and porters try to put a version >>number in their binaries that can be called with -v, -V, -h or -?. It >If typing "man" is "too cryptic", ln it to "help" on your system; if using >"man" is too cryptic, archie for "help"... there are many excellent packages. >To exaggerate a [very] little, every third Senior project is a help system. It also might be an idea to support the way GNU is doing it with their programs these days (which I believe is POSIX motivated)---with the --version flag. I would guess that this wouldn't break any current flags, and could end up being rather benign to add..... Toodlepip! Marc 'em.