HELMER%SDNET.BITNET@vm1.nodak.edu (Guy Helmer) (09/23/89)
Does anyone have C-News running on Minix? I'm (finally) getting UUPC compiled and I'm ready to get C-News up, so if anyone has patches or hints, please send mail. Thanks. -- Guy Helmer AT&T: (605) 256-5315 Dakota State University Computing Services (605) 256-6411 BITNET: HELMER@SDNET
housel@en.ecn.purdue.edu (Peter S. Housel) (09/23/89)
In article <24690@louie.udel.EDU>, HELMER%SDNET (Guy Helmer) writes: >Does anyone have C-News running on Minix? I'm (finally) getting UUPC >compiled and I'm ready to get C-News up, so if anyone has patches or >hints, please send mail. Thanks. >-- Guy Helmer AT&T: (605) 256-5315 I'm working REALLY hard on it, but it's still not there yet. Notable problems, in no particular order: 1) AWK. C News makes extensive use of AWK, and Minix doesn't really have an acceptable version. I have ported GNU Awk version 1.02 (the "old AWK" version) to Minix, and plan to post it Real Soon Now. Note that it will require the floating point package. 2) fopen(file, "[rwa]+") gets used, and the old stdio can't handle it. Earl Chew's stdio package should take care of this. 3) expr's colon operator gets used. (C news was one of the reasons my replacement version got written.) There are minor problems with other programs that mostly get used in shell scripts. 4) sh. This is the biggest problem of all. C news is largely written in /bin/sh scripts, and the Minix version of sh doesn't hold together very well under stress. Larry Wall's 'Configure' script and the C news programs show this very well. In particular, '$'-variable expansion and '\'-quoting don't work according to spec at all. The "." (source) command is a disaster. I have fixes for some of the problems, but they're not ready yet. 5) ".o" files. This is a problem with porting almost anything to Minix. Somebody, please, write an 'as' and 'ld'! (Preferably with simulated virtual memory...) 6) memory. Since everything is written in 'sh'-scripts, you can easily get things nested deeply enough that a system with small memory will run out of memory pretty quickly - especially in a system without the Protected Memory extensions. 7) The version of 'patch' distributed with Minix had problems coping with some of the C news updates. Upgrading from Patch 2.0 patchlevel 4 to patchlevel 12 solves this problem. 8) expire makes minor use of floating point for date conversions. In short: it's a challenge, some help would be appreciated, and as soon as it get done you will know about it. (And then we can start flooding news.newsites :-) -Peter S. Housel- housel@ecn.purdue.edu ...!pur-ee!housel
ast@cs.vu.nl (Andy Tanenbaum) (09/23/89)
In article <15809@ea.ecn.purdue.edu> housel@en.ecn.purdue.edu (Peter S. Housel) writes: > >1) AWK. C News makes extensive use of AWK, and Minix doesn't really >have an acceptable version. I have ported GNU Awk version 1.02 (the I now have bawk running. I was going to wait until 1.4b before posting it, but I will do it now. Be wary of GNU because P-H will never allow that in the distribution without a formal, legal contract from the copyright owner, FSF, which they will never accept. >2) fopen(file, "[rwa]+") gets used, and the old stdio can't handle it. >Earl Chew's stdio package should take care of this. Try to be careful. Although I will almost certainly replace stdio for 2.0, it is not yet clear which I will use. There are several candidates. >5) ".o" files. This is a problem with porting almost anything to Minix. >Somebody, please, write an 'as' and 'ld'! We have as and ld here. It is a matter of finding time to port them. I have hopes of getting them into 2.0 >7) The version of 'patch' distributed with Minix had problems coping >with some of the C news updates. Upgrading from Patch 2.0 patchlevel 4 >to patchlevel 12 solves this problem. Could you post the upgrades for patch relative to 1.3 please? Andy Tanenbaum (ast@cs.vu.nl)
cechew@bruce.OZ (Earl Chew) (09/25/89)
From article <3349@ast.cs.vu.nl>, by ast@cs.vu.nl (Andy Tanenbaum): > In article <15809@ea.ecn.purdue.edu> housel@en.ecn.purdue.edu (Peter S. Housel) writes: >>2) fopen(file, "[rwa]+") gets used, and the old stdio can't handle it. >>Earl Chew's stdio package should take care of this. > Try to be careful. Although I will almost certainly replace stdio for 2.0, > it is not yet clear which I will use. There are several candidates. It is not unreasonable to expect that if the use of the stdio does not rely on `features' of a particular implementation, then it should work regardless of which stdio is eventually adopted. Note, that even with my stdio, a and a+ will not work properly when several processes attempt to append simultaneously. There are some small problems with the current version of my stdio. I have changed it now to try to make it Posix/ANSI conformant. I haven't had much feedback from the news group on the package. I presume those of you (yes, both of you :-) who are using it haven't hit the bugs yet :-). Earl
cechew@bruce.OZ (Earl Chew) (09/25/89)
From article <1547@bruce.OZ>, by cechew@bruce.OZ (Earl Chew): > Note, that even with my stdio, a and a+ will not work properly when several > processes attempt to append simultaneously. Oops, what I meant to write was that there is no guarantee that a and a+ will work properly when several processes attempt to append to the same file. This problem should go away when Simon Poole's patches go in (3 argument opens). Earl
fischer@iesd.auc.dk (Lars P. Fischer) (09/26/89)
In article <3349@ast.cs.vu.nl> ast@cs.vu.nl (Andy Tanenbaum) writes: >In article <15809@ea.ecn.purdue.edu> housel@en.ecn.purdue.edu (Peter S. Housel) writes: >> >>1) AWK. C News makes extensive use of AWK, and Minix doesn't really >>have an acceptable version. I have ported GNU Awk version 1.02 (the >I now have bawk running. I was going to wait until 1.4b before posting it, >but I will do it now. Be wary of GNU because P-H will never allow that >in the distribution without a formal, legal contract from the copyright owner, >FSF, which they will never accept. Which should not keep *us* from using GNU products. GNU products are in general of a very high quality, and GNU AWK is a very fine product indeed. If P-H bails out near GNU AWK (:-), shame on them! Its really silly if the GNU and MINIX efforts are going to work against each other, instead of cooperating (yes, I know that the RMS and AST disagree in many ways, not the least of which is the proper amount of memory to use for a given task :-). /Lars -- Copyright 1989 Lars Fischer; you can redistribute only if your recipients can. Lars Fischer, fischer@iesd.auc.dk, {...}!mcvax!iesd!fischer Department of Computer Science, University of Aalborg, DENMARK. Why is it that some people insists on reinventing Square Wheels? - Mike O'Dell (on netnews)
Leisner.Henr@xerox.com (marty) (09/26/89)
Lars Fisher (fischer@iesd.auc.dk) sez: [ Which should not keep *us* from using GNU products. GNU products are in general of a very high quality, and GNU AWK is a very fine product indeed. If P-H bails out near GNU AWK (:-), shame on them! Its really silly if the GNU and MINIX efforts are going to work against each other, instead of cooperating (yes, I know that the RMS and AST disagree in many ways, not the least of which is the proper amount of memory to use for a given task :-). ] I agree totally with Lars. I don't understand what the problem is with P-H -- the GPL (Gnu Public License) seems clear enough as far as distribution requirements go. The GNU tools tend to be bigger than the equivalent Minix tools but they work and they're of exceedingly high quality (I use them on Sun386i, Sun/4 and the ones that fit under Ms/Dos). I can't believe there isn't a reasonable way for P-H and FSF to work something out. I remember the initial version of Bawk many years ago -- I couldn't get it to do anything. I really think the FSF tools would enhance Minix as a system. The alternative is going to be a plethora of versions of programs (which I'm beginning to see). A large percentage of GNU tools come with excellent documentation (I learned awk from the gawk manual, the gnu make document is great). But there's a price -- 100 page manuals in machine readable format take up space. marty ARPA: leisner.henr@xerox.com GV: leisner.henr NS: leisner:wbst139:xerox UUCP: hplabs!arisia!leisner
ast@cs.vu.nl (Andy Tanenbaum) (09/27/89)
In article <24898@louie.udel.EDU> Leisner.Henr@xerox.com (marty) writes: >I agree totally with Lars. I don't understand what the problem is with P-H >-- the GPL (Gnu Public License) seems clear enough as far as distribution >requirements go. P-H has real lawyers, not amateur net lawyers. When they see that something is copyrighted by a third party, they want a clear legal release that will stand up in court in the event that they get sued for copyright infringement. Courts just don't understand funny notices in programs. P-H has a form, devised by its lawyers, which all of its authors are required to use when they use copyrighted material. In most cases, the author sends the form to the copyright owner, who either signs it and sends it back, or says that it is prepared to sign it upon payment of some fee. From what I know of FSF they are not about to sign a form devised by P-H's lawyers. On the other hand, we just had a discussion about "Supplementary disks". I would certainly have no objection to somebody porting the GNU software to MINIX and preparing the appropriate number of disks containing them as MINIX file systems, for example (or tar, or whatever). As long as the person or organization distributing them obeyed the GNU rules and was prepared to incur the legal risk, that would seem to be ok. I will continue to collect software that I write, PD software, and copyrighted software from others from whom the legal authorization is clear. There is a MINIX Centre in the U.K. that might do this for Europe if they want to, and perhaps a similar thing can be set up in the U.S. There is, incidentally, considerable demand for access to comp.os.minix from people who have modems but are not on USENET. Such a cencould provide that access by storing comp.os.minix on a disk accessible via FIDO, etc., possibly charging for access at some reasonable rate. Comments? Andy Tanenbaum (ast@cs.vu.nl)
chasm@attctc.Dallas.TX.US (Charles Marslett) (09/27/89)
In article <24898@louie.udel.EDU>, Leisner.Henr@xerox.com (marty) writes: > Lars Fisher (fischer@iesd.auc.dk) sez: > [ > Which should not keep *us* from using GNU products. GNU products are > in general of a very high quality, and GNU AWK is a very fine product > indeed. If P-H bails out near GNU AWK (:-), shame on them! This is true, or at least I believe it to be, don't some companies (NeXT? Sun?) ship some FSF software with their systems? As I understand it, the public license prohibits P-H from using GCC or BISON to build software not distributed in source form (that is, the C compiler, unless you consider the ACK output to be source ;^), but the sources to emacs (yuk, yuk, yuk), AWK and so forth should be no problem. P-H would just have to disclaim copyright control over that part of the distribution -- say, put it on an seperate set of diskettes to make housekeeping easier. It does add to the "bulk" problem, though. > . . . > ] > I agree totally with Lars. I don't understand what the problem is with P-H > -- the GPL (Gnu Public License) seems clear enough as far as distribution > requirements go. > > The GNU tools tend to be bigger than the equivalent Minix tools but they > work and they're of exceedingly high quality (I use them on Sun386i, Sun/4 > and the ones that fit under Ms/Dos). I would agree, if you leave out the debugger and make (can't get either to work RIGHT under 386/ix -- make refuses to do anything with the gnu-emacs make file!). > I can't believe there isn't a reasonable way for P-H and FSF to work > something out. I remember the initial version of Bawk many years ago -- I > couldn't get it to do anything. Really, I have stopped giving out any changes I make to FSF products because there is one major problem with doing so -- the junior lawyers on the circuit will distroy you if you do anything wrong (inadvertently) or if you do anything they disagree with. I posted a set of minor changes to gnu-grep early this year (for MSDOS) as a set of cdiffs, and later, in response to requests from several poeple, I posted the executable. For weeks, I was berated for not posting the full source (a deliberate decision: the patches were 300-400 lines, the source is at least 10 times that), and for not posting the COPYING file with my patches and executables (inadvertent omission). I WILL NEVER DO THAT AGAIN! The threat of jail is nowhere near as fearful as that of flamage ;^)!!! > The alternative is going to be a plethora of versions of programs (which > I'm beginning to see). Perhaps, but they may really be better. The old v7_make and its successors seem much more functional than gnumake 3.05 (and is much smaller, too). I would be glad to post it, when I find out why it doesn't run the gnuemacs makefile either! > A large percentage of GNU tools come with excellent documentation (I > learned awk from the gawk manual, the gnu make document is great). But > there's a price -- 100 page manuals in machine readable format take up > space. And take my comments about gnumake with a grain of salt -- I am really hard on some programs, and so is Stallman's emacs! > marty > ARPA: leisner.henr@xerox.com > GV: leisner.henr > NS: leisner:wbst139:xerox > UUCP: hplabs!arisia!leisner =========================================================================== Charles Marslett STB Systems, Inc. <== Apply all standard disclaimers Wordmark Systems <== No disclaimers required -- that's just me chasm@attctc.dallas.tx.us