From: UnixOS2 Archive To: "UnixOS2 Archive" Date: Fri, 5 Apr 2002 04:23:23 EST-10EDT,10,-1,0,7200,3,-1,0,7200,3600 Subject: [UnixOS2_Archive] No. 185 ************************************************** Thursday 04 April 2002 Number 185 ************************************************** Subjects for today 1 Re: Gnat 3.14p OS/2 gnat.adc file problem : Vincent Marciante 2 Re: unixos2 check : csaba.raduly at sophos.com 3 SLANG curses programs : John Poltorak 4 Re: iconv.a : John Poltorak 5 Re: unixos2 check : John Poltorak 6 Re: unixos2 check : John Poltorak 7 Default value of IFS : John Poltorak 8 Make 3.79.1 problem with FIND : John Poltorak 9 New diffutils : John Poltorak 10 Re: Texinfo v4.1 : John Poltorak 11 Re: Boost. Battle II. : csaba.raduly at sophos.com 12 PHP-Nuke : John Poltorak 13 Re: unixos2 check : Stefan Neis 14 Re: PHP-Nuke : Adrian Gschwend" 15 Re: Make 3.79.1 problem with FIND : Andreas Buening 16 Re: unixos2 check : Andreas Buening **= Email 1 ==========================** Date: Fri, 05 Apr 2002 00:11:00 -0500 From: Vincent Marciante Subject: Re: Gnat 3.14p OS/2 gnat.adc file problem Vincent Marciante wrote: > > I'm installing eCS 1.0 to see if there is any difference. > I still get the same result. Are the following the correct files to be using? [D:\user\vinny\GNAT\scratch]dir f:\emx*,f:\gnu*,f:\gnat* The volume label in drive F is ECOMSTATION. The Volume Serial Number is A801:4814. Directory of F:\ 4-04-02 10:51p 0 emx 11-03-99 12:24a 1360246 267 emxdev1.zip 11-03-99 12:15a 195407 267 emxdev2.zip 3-21-02 2:34a 554166 258 emxrt_09d04.zip 4 file(s) 2109819 bytes used 11-03-99 12:24a 1075274 267 gnudev1.zip 11-03-99 12:28a 732747 267 gnudev2.zip 2 file(s) 1808021 bytes used 3-05-02 8:12p 6268133 299 gnat-3_14p-os2-bin-20020303.zip 1 file(s) 6268133 bytes used 650646016 bytes free [D:\user\vinny\GNAT\scratch] **= Email 2 ==========================** Date: Fri, 5 Apr 2002 09:56:57 +0100 From: csaba.raduly at sophos.com Subject: Re: unixos2 check On 04/04/2002 21:49:34 Andreas Buening wrote: >John Poltorak wrote: [snip announcement of unixos2-check] >> >> Maybe it could be incorporated into the UnixOS/2 Autoconf package so it >> would automatically be invoked when running Autoconf on OS/2 and would >> verify that the build environment was correct at run time... > >No, definitely not. >a) You don't want a complete system check if you install any >mini package >b) the script may become loooong on the long term which >means slooow. >c) The idea is that you run this script only once to make >your build environment "compatible". After that you don't need >it. > It might come in handy if change your include paths, your CONFIG.SITE, install new shell, etc. You mean you never change your system at all ? :-) -- Csaba Ráduly, Software Engineer Sophos Anti-Virus email: csaba.raduly at sophos.com http://www.sophos.com US Support: +1 888 SOPHOS 9 UK Support: +44 1235 559933 **= Email 3 ==========================** Date: Fri, 5 Apr 2002 10:01:51 +0000 From: John Poltorak Subject: SLANG curses programs When trying to compile the sample curses programs included with SLANG I still get errors such as these:- rain.c:23 (C:\TMP\ccc59597): Undefined symbol _SLcurses_initscr referenced from text segment rain.c:24 (C:\TMP\ccc59597): Undefined symbol _SLcurses_nil referenced from text segment rain.c:25 (C:\TMP\ccc59597): Undefined symbol _SLcurses_nil referenced from text segment rain.c:36 (C:\TMP\ccc59597): Undefined symbol _SLcurses_Stdscr referenced from text segment rain.c:36 (C:\TMP\ccc59597): Undefined symbol _SLcurses_wmove referenced from text segment rain.c:36 (C:\TMP\ccc59597): Undefined symbol _SLcurses_Stdscr referenced from text segment rain.c:36 (C:\TMP\ccc59597): Undefined symbol _SLcurses_waddch referenced from text segment Has anyone ever managed to build them? -- John **= Email 4 ==========================** Date: Fri, 5 Apr 2002 10:14:13 +0000 From: John Poltorak Subject: Re: iconv.a On Thu, Apr 04, 2002 at 09:49:42PM +0100, Andreas Buening wrote: > Holger Veit wrote: > > [_nl_msg_cat_cntr] > > To close this endless thread Unfortunately, this thread is far from closed since it still seems impossible to build GETTEXT, at least for me. > I've looked into the gettext > sources. gettext 0.10.35 had a "int _nl_msg_cat_cntr = 0;" > statement which seems to have been removed later. At least > gettext 0.10.39 and above simply use "int _nl_msg_cat_cntr;" > I don't know why. However, I had no problems to compile > gettext 0.10.39 with "emxexp -u" and gcc 2.95, so emxbind > can't be the only reason for this. Have you looked at 0.11.1 ? > bye, > Andreas -- John **= Email 5 ==========================** Date: Fri, 5 Apr 2002 10:30:25 +0000 From: John Poltorak Subject: Re: unixos2 check On Fri, Apr 05, 2002 at 09:56:57AM +0100, csaba.raduly at sophos.com wrote: > > On 04/04/2002 21:49:34 Andreas Buening wrote: > > >John Poltorak wrote: > [snip announcement of unixos2-check] > >> > >> Maybe it could be incorporated into the UnixOS/2 Autoconf package so it > >> would automatically be invoked when running Autoconf on OS/2 and would > >> verify that the build environment was correct at run time... > > > >No, definitely not. > >a) You don't want a complete system check if you install any > >mini package > >b) the script may become loooong on the long term which > >means slooow. > >c) The idea is that you run this script only once to make > >your build environment "compatible". After that you don't need > >it. > > > > It might come in handy if change your include paths, your CONFIG.SITE, > install new shell, etc. > You mean you never change your system at all ? :-) One of the biggest problems I've found is the difficulty in finding a stable environment. My 'Unix build system' gets changed several times a day sometimes in an attempt to coersce some programs into working. This is clearly a ridiculous situation, but hopefully this can be corrected once some key building blocks are in place, the most important of which is gcc 3.0.3, but that in itself require a lot of other changes... -- John **= Email 6 ==========================** Date: Fri, 5 Apr 2002 10:46:52 +0000 From: John Poltorak Subject: Re: unixos2 check On Thu, Apr 04, 2002 at 09:49:34PM +0100, Andreas Buening wrote: > John Poltorak wrote: > [snip] > > > > > The unixos2 check script would probably work best with a config.site file > > > > which included things like:- > > > > > > > > export ac_executable_extensions=".exe" > > > > > > If you haven't set that variable the script will terminate. > > > > I'm not sure how many people would know how to set it. By including a > > default config.site it may work automagically > > The script tests how a "real" configure script would see > your environment, not how it should see it. Therefore if > you don't have CONFIG_SITE and there's no config.site in > your prefix tree then there's no config.site to load. I'm not sure how many OS/2 users have ever come across a config.site file... IMV it's something we should try and encourage. > > > > > I would suggest that a sample config.site file was included in a > > > > subsequent release of your script. It looks like a useful check that could > > > > save many wasted hours over time. > > > > > > It's a "normal" autoconf script. This means it will read > > > you config.site from $CONFIG_SITE or from your prefix dir (!). > > > > Yes, but I'm suggesting that we should try devising a standard config.site > > file for use in the building of UnixOS/2 packages and include it as a > > sample with your unixos2 checker. > > Yes, if you have one. ;-) I guess you have to start somewhere. Here are some entries I think it should have:- #! /bin/sh # # config.site # UXRT=${UNIXROOT-c:} SHELL="$UXRT/bin/sh" CC=gcc CXX=gcc AWK=awk LEX=flex RANLIB=echo export PATH=`echo -E $PATH | tr '\\\\' '/'` export ac_executable_extensions=".exe" prefix=$UXRT/usr/local Additional suggestions would be welcomed... I'm still not sure when 'export' needs to be included if a variable is set. > bye, > Andreas -- John **= Email 7 ==========================** Date: Fri, 5 Apr 2002 10:52:34 +0000 From: John Poltorak Subject: Default value of IFS What is the default value of IFS? Can I reset it to the default value in a shell script by? :- IFS= -- John **= Email 8 ==========================** Date: Fri, 5 Apr 2002 13:24:17 +0000 From: John Poltorak Subject: Make 3.79.1 problem with FIND When using Make v3.79.1 with the Makefile.os2 included here:- http://hobbes.nmsu.edu/pub/os2/util/disk/gnufind.zip I get:- make -f Makefile.os2 dll find xargs locate \ CC="gcc -O -Zomf" AR="emxomfar" RANLIB="emxomfar s" O=".obj" A=".lib" \ LFLAGS="-s -Zcrtdll -Zstack 512" LIBS="../lib/find.lib ../lib/regex.lib" make: *** empty string invalid as file name. Stop. make: *** [release] Error 2 This does not occur when using v3.76.1. -- John **= Email 9 ==========================** Date: Fri, 5 Apr 2002 13:54:05 +0000 From: John Poltorak Subject: New diffutils After many years since the last update there is a new release of the diff utils. See:- ftp://ftp.gnu.org/pub/gnu/diffutils/diffutils-2.8.tar.gz It would be nice to have this build on OS/2. Has anyone tried? According to NEWS:- User-visible changes in version 2.8: * cmp and diff now conform to POSIX 1003.1-2001 (IEEE Std 1003.1-2001) if the underlying system conforms to POSIX and if the _POSIX2_VERSION environment variable is set to 200112. Conformance removes support for `diff -NUM', where NUM is a number. Use -C NUM or -U NUM instead. * cmp now supports trailing operands SKIP1 and SKIP2, like BSD cmp. * cmp -i or --ignore-initial now accepts SKIP1:SKIP2 option value. * New cmp option: -n or --bytes. * cmp's old -c or --print-chars option has been renamed; use -b or --print-bytes instead. * cmp now outputs "byte" rather than "char" outside the POSIX locale. * cmp -l's index column width now adjusts to fit larger (or smaller) files. * cmp -l -s and cmp -s -l are not allowed. Use cmp -s or cmp -l instead. * diff uses ISO 8601 style time stamps for output times (e.g. "2001-11-23 16:44:36.875702460 -0800") unless in the C or POSIX locale and the -c style is specified. * diff's -I and -F options use the regexp syntax of grep, not of Emacs. * diff now accepts multiple context arguments, and uses their maximum value. * New diff and sdiff options: -E --ignore-tab-expansion --strip-trailing-cr * New diff options: --from-file=FILE, --to-file=FILE --ignore-file-name-case --no-ignore-file-name-case * New diff3 and sdiff option: --diff-program=PROGRAM * The following diff options are still accepted, but are no longer documented. They may be withdrawn in future releases. -h (omit; it has no effect) -H (use --speed-large-files instead) -L (use --label instead) -P (use --unidirectional-new-file instead) --inhibit-hunk-merge (omit; it has no effect) * Recursive diffs now sort file names according to the LC_COLLATE locale category if possible, instead of using native byte comparison. * Diff printf specs can now use the "0" and "'" flags. * The new sdiff interactive command `ed' precedes each version with a header. * On 64-bit hosts, files larger than 2 GB can be compared. * Some internationalization support has been added, but multibyte locales are still not completely supported yet. * Some diagnostics have been reworded slightly for consistency. Also, `diff -D FOO' now outputs `/* ! FOO */' instead of `/* not FOO */'. * The `patch' part of the manual now describes `patch' version 2.5.4. * Man pages are now distributed and installed. * There is support for DJGPP; see the 'ms' subdirectory and the files m4/dos.m4 and */setmode.*. -- John **= Email 10 ==========================** Date: Fri, 5 Apr 2002 14:09:15 +0000 From: John Poltorak Subject: Re: Texinfo v4.1 On Wed, Mar 20, 2002 at 01:06:06AM +0900, Jun Sawataishi wrote: > I just put a patch kit in my web directory. > > http://homepage1.nifty.com/jsawa/gnu/texinfo-4.1-OS2-patch.zip > John, If you have spare time to test and to make a package for > newly built texinfo, please upload it in UNIXOS2 ftp site. I have built it a number of times but get inconsistant results, however, I've just notice that v4.2 has been released. See:- ftp://ftp.gnu.org/pub/gnu/texinfo/texinfo-4.2.tar.gz I don't know if this is a bug fix version... Will your patch kit work with this release? > # SAWATAISHI Jun -- John **= Email 11 ==========================** Date: Fri, 5 Apr 2002 14:56:38 +0100 From: csaba.raduly at sophos.com Subject: Re: Boost. Battle II. On 27/03/2002 07:52:50 owner-os2-unix wrote: >Stephen Amadei wrote: >> >> I noticed the GCC V.3 libraries put alot of files in g++-v3 that >> look like .h files, but they have no extension. What's with that? > >Dropping the .h is ANSI-C++ standard, e.g. see >http://www.cplusplus.com/doc/ansi/hfiles.html. > Then they are the ones who should be beaten :-) Why did they have to break a perfectly good naming convention ? -- Csaba Ráduly, Software Engineer Sophos Anti-Virus email: csaba.raduly at sophos.com http://www.sophos.com US Support: +1 888 SOPHOS 9 UK Support: +44 1235 559933 **= Email 12 ==========================** Date: Fri, 5 Apr 2002 16:13:57 +0000 From: John Poltorak Subject: PHP-Nuke I had never heard of PHP-Nuke before today, but apparently it works on OS/2 which according to the home page is a supported platform. See:- http://www.phpnuke.org Now, to find out what it actually does... Has anyone used it? Is it worth looking at? -- John **= Email 13 ==========================** Date: Fri, 5 Apr 2002 17:49:00 +0200 (CEST) From: Stefan Neis Subject: Re: unixos2 check On Fri, 5 Apr 2002 csaba.raduly at sophos.com wrote: > > On 04/04/2002 21:49:34 Andreas Buening wrote: > > >John Poltorak wrote: > [snip announcement of unixos2-check] > >> > >> Maybe it could be incorporated into the UnixOS/2 Autoconf package so it > >> would automatically be invoked when running Autoconf on OS/2 and would > >> verify that the build environment was correct at run time... > > > >your build environment "compatible". After that you don't need > >it. > > > > It might come in handy if change your include paths, your CONFIG.SITE, > install new shell, etc. > You mean you never change your system at all ? :-) You mean you change your system for every build? :-) After changing, it's still easier to call that verification script manually than havin each configure script spent minutes on tests which are needed in maybe 5% of the cases ... Regards, Stefan **= Email 14 ==========================** Date: Fri, 05 Apr 2002 18:10:53 +0200 (CDT) From: "Adrian Gschwend" Subject: Re: PHP-Nuke On Fri, 5 Apr 2002 16:13:57 +0000, John Poltorak wrote: >Has anyone used it? Is it worth looking at? It's a forum-like software for all kind of portals. Just check the homepage and you will exactly see what it can do :-) But I guess phpnuke contains more bugs than any other php based website. They disregard almost every rule of php security people should take care of if the write php code :-))) If you check the Bugtraq Archive you will find tons of exploits for phpnuke, almost one bug per week ;-) I wouldn't recommend it that much because of this cu Adrian -- Adrian Gschwend at OS/2 Netlabs ICQ: 22419590 ktk at netlabs.org ------- The OS/2 OpenSource Project: http://www.netlabs.org **= Email 15 ==========================** Date: Fri, 05 Apr 2002 20:44:20 +0100 From: Andreas Buening Subject: Re: Make 3.79.1 problem with FIND John Poltorak wrote: > > When using Make v3.79.1 with the Makefile.os2 included here:- > > http://hobbes.nmsu.edu/pub/os2/util/disk/gnufind.zip [error message] Confirmed. I'll see how this can be fixed. bye, Andreas -- One OS to rule them all, One OS to find them, One OS to bring them all and in the darkness bind them In the Land of Redmond where the Shadows lie. **= Email 16 ==========================** Date: Fri, 05 Apr 2002 20:44:34 +0100 From: Andreas Buening Subject: Re: unixos2 check John Poltorak wrote: [config.site] > I guess you have to start somewhere. Here are some entries I think it > should have:- > > #! /bin/sh redundant > # > # config.site > # > UXRT=${UNIXROOT-c:} c: might be a FAT partition or might not exist at all > SHELL="$UXRT/bin/sh" c:/bin/sh may not exist, the user may have already copied a working shell into /bin of his build drive (e:), most scripts will still use /bin/sh. > # > # config.site > # > CC=gcc Some people may have more than one gcc installed and may want to use CC=d:/emx/bin/gcc or CC=my_pgcc, some people use CC to add extra libraries like CC="gcc -L/blurb/bla -lfoo" which is supported by autoconf (but not recommended). > CXX=gcc Will fail with C++ code if the stdcpp lib is linked. Older emx-gcc implementations don't have g++ which is required for C++. > AWK=awk the user may have only gawk, may have different awk versions. Finding awk is definitely a job for configure. > LEX=flex Same as above. > RANLIB=echo Depends. I'm not sure but I guess, ranlib is not really necessary on OS/2. > export PATH=`echo -E $PATH | tr '\\\\' '/'` Not portable. Without "-E" ksh will fail, with "-E" ash will fail. > export ac_executable_extensions=".exe" Ack. > prefix=$UXRT/usr/local If somebody runs "./configure" and kde gets installed on his FAT drive c: he will be surprised. ;-) > Additional suggestions would be welcomed... Unfortunately, it's not that easy. It's no problem if you adjust your config.site to your system, but there's not much you can blindly copy&paste. > I'm still not sure when 'export' needs to be included if a variable is > set. "export" means that child processes inherit this variable. bye, Andreas -- One OS to rule them all, One OS to find them, One OS to bring them all and in the darkness bind them In the Land of Redmond where the Shadows lie.