From: UnixOS2 Archive To: "UnixOS2 Archive" Date: Mon, 4 Nov 2002 04:40:36 EST-10EDT,10,-1,0,7200,3,-1,0,7200,3600 Subject: [UnixOS2_Archive] No. 362 ************************************************** Sunday 03 November 2002 Number 362 ************************************************** Subjects for today 1 Re: gcc 3.0.4 : Mikulas Patocka 2 Re: buildin perl 5.8: signals : Jeff Robinson 3 buildin perl 5.8: signals : Yuri Dario" 4 Re: buildin perl 5.8: signals : John Poltorak 5 Re: buildin perl 5.8: signals : Yuri Dario" 6 Newbie attemp to build perl... : Antonio Manara Rossini" 7 Should the default shell for unixos2 protect/modify backslashes in the PATH ? : Thomas Hoffmann 8 installpkg : Andreas Buening 9 Re: Should the default shell for unixos2 protect/modify backslashes in : Andreas Buening 10 Re: buildin perl 5.8: signals : Yuri Dario" **= Email 1 ==========================** Date: Mon, 4 Nov 2002 15:05:11 +0100 (CET) From: Mikulas Patocka Subject: Re: gcc 3.0.4 > > > > __eh_frame_init anywhere. > > > > > > > > I have EMX 0.9c, which version do you have? > > > > > > > > Fix: > > > > > > > > - install emx 0.9c > > > > > > > > or > > > > > > > > - copy that library c.a from emx 0.9c (might not work at all; I don't > > > > know) > > > > > > > > or > > > > > > > > - erase gcc.a and gcc.lib in gcc-3.0.4 sources, so that files from EMX > > > > will be used. Most things will work with old gcc.a from EMX, some won't > > > > (for example profiling). > > > > > > > > > > > > I will look at emx 0.9d, try to find what does this symbol mean and > > > > implement it in gcc.a. > > > > > > > > Mikulas > > > > > > > > > > Hello Mikulas, > > > > > > Thanks for the reply and for your contribution, it is very > > > welcome. > > > > > > Most EMX users, myself included, are now using EMX 0.9d with > > > GCC 2.8.1 but want to move on to a later version. > > I provided empty eh_frame_init functions, download it from > http://artax.karlin.mff.cuni.cz/~mikulas/gcc-os2/ > and try it --- I don't have emx 0.9d so I don't know. If it will work for > you, upload it on hobbes. Does this version work with emx 0.9d or not? -- Should I uload it? Mikulas > Exceptions are unused in C anyway, so I think it doesn't mattter that the > functions are empty. **= Email 2 ==========================** Date: Mon, 04 Nov 2002 17:17:24 -0600 From: Jeff Robinson Subject: Re: buildin perl 5.8: signals John Poltorak wrote: > On Mon, Nov 04, 2002 at 05:32:38PM +0100, Yuri Dario wrote: > >>Hi, >> >>I'm sorry, maybe this argument has been already discussed, but I'm new to this list. >> >>Actually I got Perl 5.8 to compile using EMX fix 0.9d: I can't use the 5.7.x binaries on Hobbes >>because they are compiled with pentiumII optimization, so they can't run on an AMD K6 >>processor (pentium class). >> >>So I managed to rebuild the recent source code: after fixing some problems in my emx >>enviroment, I got all code built. >> >>The test suite works fine on all test but those based on signal handling: near 98% of tests were >>successfull. >> >>e.g. running "perl test.pl io/pipe.t" is failing with a "Process terminated by SIGPIPE"; similar >>errors with alarm signal or the other signals. >> >>Can you suggest what could be broken? I can attach my config.sh if required. > > > We had quite a number of people comparing Perl 5.8.0 builds at one time > and almost arrived at complete pass on the tests. > > Can you post the summary you get when running > ./perl harness in the directory ./t > > > >>TIA, > > > An interesting little addition to this bit is that I compiled Perl 5.8.0 on my Slackware box just the other day. Mind you, it is version 3.6 so is using libc, not glibc... but it only successfully passed ~98% of the tests, too! Jeff -- ---------------- Whatza JamochaMUD? http://jamochamud.anecho.mb.ca Or other stuff: http://www.anecho.mb.ca/~jeffnik ----------------------------------------------------------- **= Email 3 ==========================** Date: Mon, 04 Nov 2002 17:32:38 +0100 (CET) From: "Yuri Dario" Subject: buildin perl 5.8: signals Hi, I'm sorry, maybe this argument has been already discussed, but I'm new to this list. Actually I got Perl 5.8 to compile using EMX fix 0.9d: I can't use the 5.7.x binaries on Hobbes because they are compiled with pentiumII optimization, so they can't run on an AMD K6 processor (pentium class). So I managed to rebuild the recent source code: after fixing some problems in my emx enviroment, I got all code built. The test suite works fine on all test but those based on signal handling: near 98% of tests were successfull. e.g. running "perl test.pl io/pipe.t" is failing with a "Process terminated by SIGPIPE"; similar errors with alarm signal or the other signals. Can you suggest what could be broken? I can attach my config.sh if required. TIA, **= Email 4 ==========================** Date: Mon, 4 Nov 2002 17:43:45 +0000 From: John Poltorak Subject: Re: buildin perl 5.8: signals On Mon, Nov 04, 2002 at 05:32:38PM +0100, Yuri Dario wrote: > Hi, > > I'm sorry, maybe this argument has been already discussed, but I'm new to this list. > > Actually I got Perl 5.8 to compile using EMX fix 0.9d: I can't use the 5.7.x binaries on Hobbes > because they are compiled with pentiumII optimization, so they can't run on an AMD K6 > processor (pentium class). > > So I managed to rebuild the recent source code: after fixing some problems in my emx > enviroment, I got all code built. > > The test suite works fine on all test but those based on signal handling: near 98% of tests were > successfull. > > e.g. running "perl test.pl io/pipe.t" is failing with a "Process terminated by SIGPIPE"; similar > errors with alarm signal or the other signals. > > Can you suggest what could be broken? I can attach my config.sh if required. We had quite a number of people comparing Perl 5.8.0 builds at one time and almost arrived at complete pass on the tests. Can you post the summary you get when running ./perl harness in the directory ./t > TIA, -- John **= Email 5 ==========================** Date: Mon, 04 Nov 2002 17:49:49 +0100 (CET) From: "Yuri Dario" Subject: Re: buildin perl 5.8: signals Hi, sorry, forgot to update signatures on this account, so the message was unsigned :-( Bye, Yuri Dario /* * member of TeamOS/2 - Italy * http://www.quasarbbs.net/yuri * http://www.teamos2.it * http://www.opera.com/os2/ */ **= Email 6 ==========================** Date: Mon, 04 Nov 2002 19:25:45 +0100 (CET) From: "Antonio Manara Rossini" Subject: Newbie attemp to build perl... --_=_=_=IMA.BOUNDARY.H52UIX138764=_=_=_ Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Hi, it's about a month i'm linked to this list (thanks to Mentore Siesto for addressing me here...). As a completly newbie in "porting world", i've never asked for help thinking i could do something on my own... Now i'm quite sure i could never get rid of strange problems i'm encountering everytime i try to build something! To cut a long story short: everything started when i was trying to make a porting of last MRTG, that's not so hard but it requires perl, and last version/build i've found around showed a lot of troubles (according with docu also) with fork(). SO i decided to build Perl v5.8 on my own, thinking to fix something... now i'm not so sure it was a good idea :-) 1) looking at docu on the net, ive created a unix like (i hope) structure on an free partition (drive g:\)... 2) then i set all environment strings according to emx docu and to many other scripts i've found around (look at ux2_setenv.cmd) 3) installed emx fixes and all gnu tools needed to build perl. Now, at the first test: sed "s/^/build /" build.lst | cmd /q everything ended with an "unrecoverable error". I've discovered SED doesn't wotk if something using java is running (i forgot a ziga session on background). Ok, shut down everything. Then: "A non-recoverable error occurred. The process ended. OS/2 Command Interpreter version 4.5" that's always sed.... it doesn't work!!!!! BUT sometimes, after a reboot, it works. Here it's what i could get... - - - - - - - - - [G:\unixos2\lib]ARCHIVE CFLAGS LDFLAGS PARMS MAKEPARM SRC perl-5.8.0 . ARCHIVE perl-5.8.0 CFLAGS LDFLAGS PARMS MAKEPARM . SRC /unixos2/workdir /unixos2/workdir/perl-5.8.0 build.sh[126]: autoconf: not found ./configure configure[73]: cmd: not found configure[73]: cmd: not found configure[73]: tr: not found configure[73]: tr: not found (I see you are using the Korn shell. Some ksh's blow up on configure, mainly on older exotic systems. If yours does, try the Bourne shell instead.) Process terminated by SIGSEGV Process terminated by SIGSEGV configure: Fatal Error: I can't find a Bourne Shell anywhere. Usually it's in /bin/sh. How did you even get this far? Please contact me (Perl Maintainers) at perlbug at perl.org and we'll try to straighten this all out. cp: unixos2/lib/mkinstalldirs: No such file or directory make . make: Nothing to be done for `.'. end - - - - - - - - - well, autoconf is here, sh also... Now i "suppose" O:-) there's something wrong around, but i'm not eable to find what. I'm really disappointed by my work, as i'm not eable to find a way to fix those troubles! I hope someone here could give me a little help.... Many thanks for all! P.S. Do you think it could be too hard to distribute an archive of a working environment (with all fixes, scripts,...) with a bootstrap cmd with all settings...? --- Antonio Manara Rossini Quality Manager (ISO9000) at Manara s.r.l. Edil Constructions, Italy. WSiTE: www.vrglabs.com - FiDONET: 2:332/807.14 - iCQ: 14586599 PGP 5.0i key available. --_=_=_=IMA.BOUNDARY.H52UIX138764=_=_=_ Content-Type: application/octet-stream; name="ux2_setenv.cmd" Content-Transfer-Encoding: base64 QGVjaG8gb2ZmDQpzZXQgYmxkcnQ9ZzoNCnNldCB1eHJ0PWc6DQpzZXQgb3NydD1nOg0Kc2V0IGJs ZF9ob21lPXVuaXhvczINCnNldCB1bml4cm9vdD1nOg0Kc2V0IGhvbWU9ZzovaG9tZQ0Kc2V0IHRt cD1nOi90bXANCg0KUkVNICMjIyMjIyMjIyMjIyMjIyMjIyMjIyMjIyMjIyMjIyMjIyMNClJFTSAj IyMgU0VULVNUQVRFTUVOVFMgRk9SIEVNWCBHQ0MgIyMjDQpSRU0gIyMjIyMjIyMjIyMjIyMjIyMj IyMjIyMjIyMjIyMjIyMjIw0KU0VUIFRFUk09YW5zaQ0KU0VUIFRFUk1DQVA9ZzovZXRjL3Rlcm1j YXAuZGF0DQpTRVQgVEVSTUlORk89ZzpcdXNyXGxvY2FsXHNoYXJlXHRlcm1pbmZvDQpSRU0gLS0t IFNUQVJUIG9mIEdDQyBPTkxZIFNFQ1RJT04gLS0tDQpTRVQgQ19JTkNMVURFX1BBVEg9RzovRU1Y L0lOQ0xVREUNClNFVCBMSUJSQVJZX1BBVEg9RzovRU1YL0xJQjsNClNFVCBDUExVU19JTkNMVURF X1BBVEg9RzovRU1YL0lOQ0xVREUvQ1BQO0c6L0VNWC9JTkNMVURFOy47DQpTRVQgUFJPVE9ESVI9 RzovRU1YL0lOQ0xVREUvQ1BQL0dFTjsNClNFVCBHQ0NMT0FEPTUNClNFVCBHQ0NPUFRTPS1QSVBF DQpTRVQgRU1YQk9PSz1FTVhERVYuSU5GK0VNWExJQi5JTkYrRU1YUlQuSU5GK0VNWEdOVS5JTkYN ClNFVCBIRUxQTkRYPWVwbWt3aGxwLm5keCsNClNFVCBFTVhTSEVMTD1zaA0KUkVNIC0tLSBFTkQg b2YgR0NDIE9OTFkgU0VDVElPTiAtLS0NClNFVCBFTVhPUFQ9LWMgLWgxMjANClJFTSAvLy8gR05V IFBBQ0sNClNFVCBTSEVMTD1nOlx1c3JcYmluXHNoLmV4ZQ0KU0VUIENPTkZJR19TSEVMTD1nOlx1 c3JcYmluXHNoLmV4ZQ0KU0VUIEJBU0hfU1RBUlRVUD1nOlxob21lXC5iYXNob3MyDQpTRVQgWUFD Qz1iaXNvbiAteQ0KU0VUIEFXSz1nYXdrDQpTRVQgTEVYPWZsZXgNClNFVCBBQ19NQUNST0RJUj1n Olx1c3JcYmluDQpTRVQgTTQ9ZzpcZW14XEdOVVxtNA0KU0VUIFJBTkxJQj10b3VjaA0KDQpzZXQg YmxkX2hvbWU9JWJsZHJ0JS8lYmxkX2hvbWUlDQpzZXQgYmVnaW5saWJwYXRoPS5cLjsldXhydCVc ZW14XGRsbDsldXhydCVcdXNyXGxpYjsNCnNldCByZXBvc2l0b3J5PSVibGRfaG9tZSVcYXJjaGl2 ZXNcc291cmNlDQo= --_=_=_=IMA.BOUNDARY.H52UIX138764=_=_=_-- **= Email 7 ==========================** Date: Mon, 04 Nov 2002 22:52:56 +0100 From: Thomas Hoffmann Subject: Should the default shell for unixos2 protect/modify backslashes in the PATH ? I tried to run a configure script using the "ksh 2" shell: [zappa|~/work/R/rwork151]ls -l `which sh` -rwxrwx--a 163844 Feb 1 2002 d:\bin\sh.exe This way I noticed that sh echoes an OS/2 PATH w/ backslashes verbatim and so eats some backslashes and the following letter as control characters: # echo $PATH D:\BIN;D:\USR\BIN;.;D:pps ext bash on the other hand seems to protect the output: (ash does the same, btw) [zappa|~/work/R/rwork151]echo $PATH D:\BIN;D:\USR\BIN;.;D:\apps\text I would think that the "standard shell" should try to protect the PATH the way (b)ash does. On the other hand I thought that "ksh 2" would end the switching/ renaming of different shells. Is there any consensus how the above should work? **= Email 8 ==========================** Date: Mon, 04 Nov 2002 23:24:37 +0100 From: Andreas Buening Subject: installpkg Hello! Is installpkd.cmd still supposed to be our package installation program? If yes, is there a maintainer of it? I couldn't find any email address. I think we'll need some upgrades to this program. 1) installation log files should be placed into /var/lib/unixos2 instead of /var/log. I guess, /var/log is for logfiles of programs like apache which you can "rm -fR" if you're low on disk space. 2) An additional keyword SCRIPTFILE that indicates that this file is a shell script whose 1st line "#!/bin/foo" has to be changed by an absolute path name "#!x:/bin/foo" at installation time. 3) Yet another additional keyword SCRIPTFILE2 that indicates that this file contains a lot of path names like "/usr/somewhere" that have to be replaced by "x:/usr/somewhere" at installation time. 4) Support for the installation of large archives (hundreds of files). The current keyword "FILE dir/file" runs "unzip ... dir/file" which is a) very slow for huge numbers of files, b) very hard to maintain because you need hundreds of those FILE entries in your PKGINFO file. Maybe something like "unzip ... * -x exclude_list" might be considerable. Any feedback is appreciated. :-) 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 Mordor where the Shadows lie. **= Email 9 ==========================** Date: Mon, 04 Nov 2002 23:43:59 +0100 From: Andreas Buening Subject: Re: Should the default shell for unixos2 protect/modify backslashes in Thomas Hoffmann wrote: > > I tried to run a configure script using the "ksh 2" shell: > > [zappa|~/work/R/rwork151]ls -l `which sh` > -rwxrwx--a 163844 Feb 1 2002 d:\bin\sh.exe > > This way I noticed that sh echoes an OS/2 PATH w/ backslashes verbatim > and so eats some backslashes and the following letter as control characters: > > # echo $PATH > D:\BIN;D:\USR\BIN;.;D:pps ext > > bash on the other hand seems to protect the output: (ash does the same, btw) > > [zappa|~/work/R/rwork151]echo $PATH > D:\BIN;D:\USR\BIN;.;D:\apps\text This is a (not really) well known issue. The bad news is that ksh seems to be right. I don't remember whether it's Posix or SysV but it's standardized. :-( Btw. that's the reason why I implemented such a '\' -> '/' feature into the OS/2 release of autoconf. > I would think that the "standard shell" should try to protect the PATH > the way (b)ash does. On the other hand I thought that "ksh 2" would end > the switching/ renaming of different shells. > > Is there any consensus how the above should work? I propose, the philosophy should be to use a config.site file for this purpose. This is better than to change the behaviour of existing shells (cause them to change '\' by default to '/' in PATH) or to expect from the user that he does this job himself. To be more exactly: A file /etc/unixroot/config.site should be part of the UnixOS/2 base package with the recommendation for all developers (i.e.: it's an optional env. var.) to set CONFIG_SITE=$UNIXROOT/etc/unixroot/config.site. (/etc because config.site is neither a "readonly" (/usr) file of the distro nor a "variable" (/var) file but a config file which can be edited by the user like XF86Config.) 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 Mordor where the Shadows lie. **= Email 10 ==========================** Date: Mon, 04 Nov 2002 23:53:15 +0100 (CET) From: "Yuri Dario" Subject: Re: buildin perl 5.8: signals Hi John, >Can you post the summary you get when running >./perl harness in the directory ./t here: 62 tests and 535 subtests skipped. Failed 19/726 test scripts, 97.38% okay. 180/68636 subtests failed, 99.74% okay. Bye, Yuri Dario /* * member of TeamOS/2 - Italy * http://www.quasarbbs.net/yuri * http://www.teamos2.it * http://www.opera.com/os2/ */