Date: Mon, 29 Nov 2004 00:13:31 EST-10EDT,10,-1,0,7200,3,-1,0,7200,3600 Subject: [UnixOS2_Archive] No. 461 ************************************************** Sunday 28 November 2004 Number 461 ************************************************** Subjects for today 1 Re: DosAllocSharedMem in libc v0.6-beta1 : Yuri Dario" 2 Netstat : John Poltorak 3 Re: PostgreSQL does build with GCC 3.3.5 : lsunley at mb.sympatico.ca 4 TVFS expertise wanted : John Poltorak 5 Does link do anything : lsunley at mb.sympatico.ca 6 Re: Does link do anything : Dave Yeo" 7 Re: Does link do anything : Yuri Dario" 8 Re: TVFS expertise wanted : Dave Yeo" 9 Re: Does link do anything : lsunley at mb.sympatico.ca 10 Re: DosAllocSharedMem in libc v0.6-beta1 : Dave Yeo" 11 Re: TVFS expertise wanted : Henry Sobotka 12 postgreSQL mods available : lsunley at mb.sympatico.ca 13 Re: TVFS expertise wanted : John Poltorak 14 Re: DosAllocSharedMem in libc v0.6-beta1 : John Poltorak 15 Re: Does link do anything : lsunley at mb.sympatico.ca 16 Re: postgreSQL mods available : lsunley at mb.sympatico.ca 17 Re: Does link do anything : Dave Yeo" 18 Re: Does link do anything : lsunley at mb.sympatico.ca 19 BSD Makefiles : John Poltorak 20 Re: TVFS expertise wanted : Dave Yeo" 21 Re: Building Python 2.3.4 with Innotek GCC 3.3.5 beta 1 : Knut Stange Osmundsen 22 Re: Does link do anything : Knut Stange Osmundsen 23 Re: Does link do anything : Knut Stange Osmundsen 24 Re: TVFS expertise wanted : Henry Sobotka 25 Re: Does link do anything : Stefan.Neis at t-online.de 26 Re: BSD Makefiles : Stefan.Neis at t-online.de 27 Re: Does link do anything : Yuri Dario" 28 Re: Building Python 2.3.4 with Innotek GCC 3.3.5 beta 1 : Andrew MacIntyre 29 Re: System V Semaphore emulation : Andrew MacIntyre 30 Re: Building Python 2.3.4 with Innotek GCC 3.3.5 beta 1 : Andrew MacIntyre 31 Re: Does link do anything : Yuri Dario" 32 Re: Building Python 2.3.4 with Innotek GCC 3.3.5 beta 1 : Paul Smedley 33 Re: Building Python 2.3.4 with Innotek GCC 3.3.5 beta 1 : John Poltorak 34 [madodel at ptdprolog.net: [VOICENWS] SW: MySQL 4.0.18 for OS/2] : John Poltorak 35 Re: Does link do anything : Stefan.Neis at t-online.de 36 Building Zope with GCC 3.3.5 : John Poltorak **= Email 1 ==========================** Date: Sat, 27 Nov 2004 14:20:50 +0100 (CET) From: "Yuri Dario" Subject: Re: DosAllocSharedMem in libc v0.6-beta1 Hi, >ate/star.html, also smake and cdrecord from the same authour ) which checks for >DosAllocSharedMem and DosSetPriority. >While these are found in EMX they are not found in libc. probably os2.h is not included, or INCL_DOS is not defined: without os2.h, functions are declared with default linkage system, that is different from _System, so linker fails. Bye, Yuri Dario /* * member of TeamOS/2 - Italy * http://www.os2power.com/yuri * http://www.teamos2.it */ **= Email 2 ==========================** Date: Sat, 27 Nov 2004 14:46:38 +0000 From: John Poltorak Subject: Netstat Having built a BSD version of ping with gcc3.3.5, I wondered if it would be possible to build something like netstat... It would be interesting to see how many MPTN programs could be replaced with open source equivalents... -- John **= Email 3 ==========================** Date: Sat, 27 Nov 2004 09:01:55 -0500 From: lsunley at mb.sympatico.ca Subject: Re: PostgreSQL does build with GCC 3.3.5 In <20041127120218.U83 at warpix.org>, on 11/27/04 at 12:02 PM, John Poltorak said: >On Fri, Nov 26, 2004 at 05:39:29PM -0500, lsunley at mb.sympatico.ca wrote: >> Just to let you all know... >> >> PostgreSQL v 8.0.0 beta 4 does build with GCC 3.3.5 b1 >This is good to know. > >> Several changes to the makefiles are required as well as some source >> changes to avoid redefinition of some things supplied in os2emx.h. >Is it possible to create a Makefile using autoconf and configure? > That's how I created the makefile set. I just ran ./configure with the --without-readline and --without-zlib and the makefile.global was created. The vanilla source tree and makefiles from the tarball require changes to set up port specific items. The various ports have conditional code based on the port name to compile port specific code and port specific commands for the make. I used ash as the shell. I built a new bison (version 1.875) as the parser requires that level of bison. I put the compiled version with all the source on Hobbes, it is built with gcc 3.2.2. >> The source code incorrectly tests for HAVE_OPTRESET rather than the >> HAVE_INT_OPTRESET the configure program sets up to test. >> >> The new setitimer stuff appears to work for the compile and build. >> >> I still have no idea if the Sys V semaphore code I wrote works properly >> yet as I have not set up all the things needed to create databases yet. >> >> I'll post an update tomorrow when I know how (and if) it all works. >Any chance of something like a build script? I'd like to see if it will >build in my environment. I'd have to zip up the entire tree (less the object files). I am going to migrate my changes to the beta 5 code this weekend. Lorne -- ----------------------------------------------------------- lsunley at mb.sympatico.ca ----------------------------------------------------------- **= Email 4 ==========================** Date: Sat, 27 Nov 2004 15:38:23 +0000 From: John Poltorak Subject: TVFS expertise wanted I've used TVFS for quite a while but only for simple tasks, but I was wondering whether it would be possible to create a mount point where I could overlay one directory over another... What I would like to do is join two directory structures both of which have a different \usr directory, but make it appear as though they are on the same TVLINKed point. Is that possible? -- John **= Email 5 ==========================** Date: Sat, 27 Nov 2004 09:47:52 -0500 From: lsunley at mb.sympatico.ca Subject: Does link do anything Hi all The unistd.h file has the prototype for link (which is used to set up a file link). Does this actually do anything? The timezone install on postgreSQL v 8 has a program called zic that sets up links for the timezone files by calling link(...) Lorne -- ----------------------------------------------------------- lsunley at mb.sympatico.ca ----------------------------------------------------------- **= Email 6 ==========================** Date: Sat, 27 Nov 2004 08:09:57 -0800 From: "Dave Yeo" Subject: Re: Does link do anything On Sat, 27 Nov 2004 09:47:52 -0500, lsunley at mb.sympatico.ca wrote: >Hi all > >The unistd.h file has the prototype for link (which is used to set up a >file link). > >Does this actually do anything? > >The timezone install on postgreSQL v 8 has a program called zic that sets >up links for the timezone files by calling link(...) Sure does. With alpha 3 I built the gnu fileutils and had working symlinks though they only work with programs compiled with libc06. Symlink data is kept in EAs and Knut was going to make them compatible with the Linux HPFS driver for this release. Tried rebuilding the file utils last night but kept getting duplicate symbol errors in libc.dll.a. Will try again after a reboot tonight. Dave **= Email 7 ==========================** Date: Sat, 27 Nov 2004 17:10:11 +0100 (CET) From: "Yuri Dario" Subject: Re: Does link do anything Hi Lorne, >The unistd.h file has the prototype for link (which is used to set up a >file link). >Does this actually do anything? if you are running libc0.6 beta1, the answer is yes: symlinks are handled using EA. BTW I have already found some bugs in handling (e.g. realpath() can enter an infinite loop in certain situations). Bye, Yuri Dario /* * member of TeamOS/2 - Italy * http://www.os2power.com/yuri * http://www.teamos2.it */ **= Email 8 ==========================** Date: Sat, 27 Nov 2004 08:20:05 -0800 From: "Dave Yeo" Subject: Re: TVFS expertise wanted On Sat, 27 Nov 2004 15:38:23 +0000, John Poltorak wrote: > >I've used TVFS for quite a while but only for simple tasks, but I was >wondering whether it would be possible to create a mount point where I >could overlay one directory over another... > >What I would like to do is join two directory structures both of which >have a different \usr directory, but make it appear as though they are on >the same TVLINKed point. Is that possible? Yes quite possible, by using the position statement in your tvlink command you can control which directory comes first. tvlink -drwnt -#999 x:\usr c:\usr tvlink -drwnt -#900 x:\usr d:\usr IIRC should be close to the right commands for C:\usr to be searched first Dave **= Email 9 ==========================** Date: Sat, 27 Nov 2004 10:36:00 -0500 From: lsunley at mb.sympatico.ca Subject: Re: Does link do anything In <200411271610.QAA365.58 at mail.warpix.org>, on 11/27/04 at 05:10 PM, "Yuri Dario" said: >Hi Lorne, >>The unistd.h file has the prototype for link (which is used to set up a >>file link). >>Does this actually do anything? >if you are running libc0.6 beta1, the answer is yes: symlinks are handled >using EA. BTW I have already found some bugs in handling (e.g. realpath() >can enter an infinite loop in certain situations). >Bye, > Yuri Dario I see, that is why nothing seems to appear for me... Lorne -- ----------------------------------------------------------- lsunley at mb.sympatico.ca ----------------------------------------------------------- **= Email 10 ==========================** Date: Sat, 27 Nov 2004 08:43:25 -0800 From: "Dave Yeo" Subject: Re: DosAllocSharedMem in libc v0.6-beta1 On Sat, 27 Nov 2004 14:20:50 +0100 (CET), Yuri Dario wrote: >Hi, > >>ate/star.html, also smake and cdrecord from the same authour ) which checks for >>DosAllocSharedMem and DosSetPriority. >>While these are found in EMX they are not found in libc. > >probably os2.h is not included, or INCL_DOS is not defined: without os2.h, functions are declared >with default linkage system, that is different from _System, so linker fails. Hmm, should of checked closer, started writing this referencing Alpha 3 which didn't have DosAllocSharedMem. Have to look closer at the configure script to see how to include os2.h. Thanx Dave **= Email 11 ==========================** Date: Sat, 27 Nov 2004 11:43:05 -0500 From: Henry Sobotka Subject: Re: TVFS expertise wanted John Poltorak wrote: > I've used TVFS for quite a while but only for simple tasks, but I was > wondering whether it would be possible to create a mount point where I > could overlay one directory over another... > > What I would like to do is join two directory structures both of which > have a different \usr directory, but make it appear as though they are on > the same TVLINKed point. Is that possible? Sure. For instance, usr1 below has a usr\foo\bar tree, usr2 usr\bar\foo. [C:\temp]tvlink -drw y:\usr c:\temp\usr1\usr [C:\temp]tvlink -drw y:\usr c:\temp\usr2\usr [C:\temp]du y:/usr 0 y:/usr/bar/foo 0 y:/usr/bar 0 y:/usr/foo/bar 0 y:/usr/foo 0 y:/usr h~ **= Email 12 ==========================** Date: Sat, 27 Nov 2004 10:37:20 -0500 From: lsunley at mb.sympatico.ca Subject: postgreSQL mods available Hi All, If anyone wants to look at this..... I have placed a zip file containing the modified makefiles , code etc for postgreSQL v8 beta 4 on my web site. The program has not been tested other than to see if the postgres.exe program displays help and configuration data (which it does). The built utilities initdb.exe etc won't run because they are looking for a postgres executable that does not have an .exe extension. The source code for semaphores is faking the presence of a signal event handler. You need the beta4 tarball from the postgresql.org web site. This file has to be unzipped over the src subdirectory of the postgreSQL build tree. http://www3.mb.sympatico.ca/~lsunley/psqlb4mods.zip -- ----------------------------------------------------------- lsunley at mb.sympatico.ca ----------------------------------------------------------- **= Email 13 ==========================** Date: Sat, 27 Nov 2004 16:49:32 +0000 From: John Poltorak Subject: Re: TVFS expertise wanted On Sat, Nov 27, 2004 at 08:20:05AM -0800, Dave Yeo wrote: > On Sat, 27 Nov 2004 15:38:23 +0000, John Poltorak wrote: > > > > >I've used TVFS for quite a while but only for simple tasks, but I was > >wondering whether it would be possible to create a mount point where I > >could overlay one directory over another... > > > >What I would like to do is join two directory structures both of which > >have a different \usr directory, but make it appear as though they are on > >the same TVLINKed point. Is that possible? > > Yes quite possible, by using the position statement in your tvlink > command you can control which directory comes first. > tvlink -drwnt -#999 x:\usr c:\usr > tvlink -drwnt -#900 x:\usr d:\usr > IIRC should be close to the right commands for C:\usr to be searched > first One of the things I would like to do is to have to separate directories for %WORKDIR% under UX2BS where one would contain files extracted from a tarball and the other would contain any newly created files. For instance c:\unixos2\workdir\perl-5.8.5 would simply contain the extracted perl-5.8.5.tar.gz d:\unixos2\workdir\perl-5.8.5 would contain any files created through the course of the build process. I'd like to be able to create a link to x:\unixos2\workdir\perl-5.8.5 but ensure the files in C: were left intact. Not sure if I have explained it well, but hopefully you will understand... I guess if I did the extract to C: first and then a RO link I should be close to what I want. > Dave -- John **= Email 14 ==========================** Date: Sat, 27 Nov 2004 16:53:48 +0000 From: John Poltorak Subject: Re: DosAllocSharedMem in libc v0.6-beta1 On Sat, Nov 27, 2004 at 08:43:25AM -0800, Dave Yeo wrote: > On Sat, 27 Nov 2004 14:20:50 +0100 (CET), Yuri Dario wrote: > > >Hi, > > > >>ate/star.html, also smake and cdrecord from the same authour ) which checks for > >>DosAllocSharedMem and DosSetPriority. > >>While these are found in EMX they are not found in libc. > > > >probably os2.h is not included, or INCL_DOS is not defined: without os2.h, functions > are declared > >with default linkage system, that is different from _System, so linker fails. > > Hmm, should of checked closer, started writing this referencing Alpha 3 which didn't > have DosAllocSharedMem. Have to look closer at the configure script to see how to > include os2.h. Maybe patching configure.in or config.h.in... Could we add a test to Autoconf to see if os2.h should be included? > Thanx Dave > -- John **= Email 15 ==========================** Date: Sat, 27 Nov 2004 10:44:20 -0500 From: lsunley at mb.sympatico.ca Subject: Re: Does link do anything In <20041127160934.A83D1B724C at generation.lgisp.net>, on 11/27/04 at 08:09 AM, "Dave Yeo" said: >On Sat, 27 Nov 2004 09:47:52 -0500, lsunley at mb.sympatico.ca wrote: >>Hi all >> >>The unistd.h file has the prototype for link (which is used to set up a >>file link). >> >>Does this actually do anything? >> >>The timezone install on postgreSQL v 8 has a program called zic that sets >>up links for the timezone files by calling link(...) >Sure does. With alpha 3 I built the gnu fileutils and had working >symlinks though they only work with programs compiled with libc06. >Symlink data is kept in EAs and Knut was going to make them compatible >with the Linux HPFS driver for this release. Tried rebuilding the file >utils last night but kept getting duplicate symbol errors in libc.dll.a. >Will try again after a reboot tonight. Dave Is that the _optarg from libc_dll.a? I got that with bison and changed it to compile and link with the -Zomf switch and the build worked. -- ----------------------------------------------------------- lsunley at mb.sympatico.ca ----------------------------------------------------------- **= Email 16 ==========================** Date: Sat, 27 Nov 2004 11:04:28 -0500 From: lsunley at mb.sympatico.ca Subject: Re: postgreSQL mods available oops... The zip file has to unzipped one directory up from the src directory... Lorne In <0I7U00G56JTKPR at l-daemon>, on 11/27/04 at 10:37 AM, lsunley at mb.sympatico.ca said: >Hi All, >If anyone wants to look at this..... >I have placed a zip file containing the modified makefiles , code etc for >postgreSQL v8 beta 4 on my web site. The program has not been tested >other than to see if the postgres.exe program displays help and >configuration data (which it does). The built utilities initdb.exe etc >won't run because they are looking for a postgres executable that does >not have an .exe extension. >The source code for semaphores is faking the presence of a signal event >handler. >You need the beta4 tarball from the postgresql.org web site. >This file has to be unzipped over the src subdirectory of the postgreSQL >build tree. >http://www3.mb.sympatico.ca/~lsunley/psqlb4mods.zip -- ----------------------------------------------------------- lsunley at mb.sympatico.ca ----------------------------------------------------------- **= Email 17 ==========================** Date: Sat, 27 Nov 2004 09:17:55 -0800 From: "Dave Yeo" Subject: Re: Does link do anything On Sat, 27 Nov 2004 10:44:20 -0500, lsunley at mb.sympatico.ca wrote: >>with the Linux HPFS driver for this release. Tried rebuilding the file >>utils last night but kept getting duplicate symbol errors in libc.dll.a. >>Will try again after a reboot tonight. Dave > >Is that the _optarg from libc_dll.a? > >I got that with bison and changed it to compile and link with the -Zomf >switch and the build worked. Yes, that was the problem with fileutils-4.1. I tried mucking about with -Zomf but had problems linking with the included libfetish.a. Seems like there should be a simple way to use -Zomf and link to included libs but what I don't know. I'm used to a.out Dave **= Email 18 ==========================** Date: Sat, 27 Nov 2004 11:37:25 -0500 From: lsunley at mb.sympatico.ca Subject: Re: Does link do anything In <20041127171741.918BBB8027 at generation.lgisp.net>, on 11/27/04 at 09:17 AM, "Dave Yeo" said: >On Sat, 27 Nov 2004 10:44:20 -0500, lsunley at mb.sympatico.ca wrote: >>>with the Linux HPFS driver for this release. Tried rebuilding the file >>>utils last night but kept getting duplicate symbol errors in libc.dll.a. >>>Will try again after a reboot tonight. Dave >> >>Is that the _optarg from libc_dll.a? >> >>I got that with bison and changed it to compile and link with the -Zomf >>switch and the build worked. >Yes, that was the problem with fileutils-4.1. I tried mucking about with >-Zomf but had problems linking with the included libfetish.a. Seems like >there should be a simple way to use -Zomf and link to included libs but >what I don't know. I'm used to a.out Dave Can't you use emxomf to convert the .a files to .lib like the emxomflibs.cmd does. The command they use in the command file is emxomf.exe -t Lorne -- ----------------------------------------------------------- lsunley at mb.sympatico.ca ----------------------------------------------------------- **= Email 19 ==========================** Date: Sat, 27 Nov 2004 18:13:58 +0000 From: John Poltorak Subject: BSD Makefiles What do I need to be able to use BSD Makefiles? I have a Makefile for SYSLOG which consists of:- # at (#)Makefile 8.1 (Berkeley) 6/6/93 # $FreeBSD: src/usr.sbin/syslogd/Makefile,v 1.3.6.3 2001/10/18 08:12:50 des Exp $ ..PATH: ${.CURDIR}/../../usr.bin/wall PROG= syslogd MAN= syslog.conf.5 syslogd.8 SRCS= syslogd.c ttymsg.c CFLAGS+=-DINET6 -I${.CURDIR}/../../usr.bin/wall WARNS?= 1 ..include What is this bsd.prog.mk ? -- John **= Email 20 ==========================** Date: Sat, 27 Nov 2004 10:24:57 -0800 From: "Dave Yeo" Subject: Re: TVFS expertise wanted On Sat, 27 Nov 2004 16:49:32 +0000, John Poltorak wrote: >On Sat, Nov 27, 2004 at 08:20:05AM -0800, Dave Yeo wrote: >> On Sat, 27 Nov 2004 15:38:23 +0000, John Poltorak wrote: >> >> > >> >I've used TVFS for quite a while but only for simple tasks, but I was >> >wondering whether it would be possible to create a mount point where I >> >could overlay one directory over another... >> > >> >What I would like to do is join two directory structures both of which >> >have a different \usr directory, but make it appear as though they are on >> >the same TVLINKed point. Is that possible? >> >> Yes quite possible, by using the position statement in your tvlink >> command you can control which directory comes first. >> tvlink -drwnt -#999 x:\usr c:\usr >> tvlink -drwnt -#900 x:\usr d:\usr >> IIRC should be close to the right commands for C:\usr to be searched >> first > >One of the things I would like to do is to have to separate directories >for %WORKDIR% under UX2BS where one would contain files extracted from a >tarball and the other would contain any newly created files. For instance > >c:\unixos2\workdir\perl-5.8.5 > >would simply contain the extracted perl-5.8.5.tar.gz > >d:\unixos2\workdir\perl-5.8.5 > >would contain any files created through the course of the build process. > >I'd like to be able to create a link to > >x:\unixos2\workdir\perl-5.8.5 > >but ensure the files in C: were left intact. > >Not sure if I have explained it well, but hopefully you will understand... > >I guess if I did the extract to C: first and then a RO link I should be >close to what I want. This might be what you are looking for. Can also make the C: readonly for extra safety tvlink -drwnt -#900 x:\xunixos2\workdir\perl-5.8.5 c:\unixos2\workdir\perl-5.8.5 tvlink -drwnt -#999 x:\xunixos2\workdir\perl-5.8.5 d:\unixos2\workdir\perl-5.8.5 Dave **= Email 21 ==========================** Date: Sat, 27 Nov 2004 19:47:46 +0100 From: Knut Stange Osmundsen Subject: Re: Building Python 2.3.4 with Innotek GCC 3.3.5 beta 1 Paul Smedley wrote: > Hi All, > I just got Python 2.3.4 to build with Knut's latest GCC. > > A few things required change: > 1) in pc/os2emx - pyconfig.h needs editing and the '#define intptr_t' > removing as it's already defined in Innotek GCC > 2) in python - thread.c - I had to comment out: > #ifdef _POSIX_THREADS > /*#include "thread_pthread.h" innote */ > #endif > as it appears that Innotek GCC defines that is has _POSIX_THREADS and > this code gets executed as well as the _OS2_THREADS when building thread.c > 3) in pc/os2emx - I had to comment out > $(EXEOPT) -aq $(PYTHON.EXE) -h$(NFILES) > & $(EXEOPT) -aq $(PYTHONPM.EXE) -h$(NFILES) > Innotek GCC's emxbind doesn't support -a - and with just -q I got an > error 'emxbind: The input and output files have the same name' > > It seems to run OK but running 'make test' I get > <...snip...> > test_fnmatch > test_fork1 > > !!!deadlock!!! LIBC Heap Recursive mutex! > > !!!deadlock!!! LIBC Heap Recursive mutex! > make: *** [test] Error 127 > > not sure if this is due to the handles not being set correctly in > python.exe or not.... This must be some kind of bug in the fmutex stuff. I've got some other error report on it too. I'll look at it next week when I get back to Germany. Kind Regards, knut **= Email 22 ==========================** Date: Sat, 27 Nov 2004 19:56:01 +0100 From: Knut Stange Osmundsen Subject: Re: Does link do anything lsunley at mb.sympatico.ca wrote: > Hi all > > The unistd.h file has the prototype for link (which is used to set up > a file link). > > Does this actually do anything? > > The timezone install on postgreSQL v 8 has a program called zic that > sets up links for the timezone files by calling link(...) link is hardlinks, i.e. not supported. The function is a stub in LIBC and will just do the parameter validation before returning with errno set to ENOSYS. Kind Regards, knut **= Email 23 ==========================** Date: Sat, 27 Nov 2004 19:52:34 +0100 From: Knut Stange Osmundsen Subject: Re: Does link do anything Yuri Dario wrote: > Hi Lorne, > > >> The unistd.h file has the prototype for link (which is used to set >> up a file link). Does this actually do anything? > > > if you are running libc0.6 beta1, the answer is yes: symlinks are > handled using EA. BTW I have already found some bugs in handling > (e.g. realpath() can enter an infinite loop in certain situations). Can you explain how to reproduce this and I'll fix it. Kind Regards, knut **= Email 24 ==========================** Date: Sat, 27 Nov 2004 14:48:57 -0500 From: Henry Sobotka Subject: Re: TVFS expertise wanted John Poltorak wrote: > > One of the things I would like to do is to have to separate directories > for %WORKDIR% under UX2BS where one would contain files extracted from a > tarball and the other would contain any newly created files. For instance > > c:\unixos2\workdir\perl-5.8.5 > > would simply contain the extracted perl-5.8.5.tar.gz > > d:\unixos2\workdir\perl-5.8.5 > > would contain any files created through the course of the build process. > > I'd like to be able to create a link to > > x:\unixos2\workdir\perl-5.8.5 > > but ensure the files in C: were left intact. > > Not sure if I have explained it well, but hopefully you will understand... Most GNU and GNUlike (e.g. mozilla, XFree86 stuff) packages support building in a directory other than the source. With them, simple TVFS linkage works fine. For instance, if k:/mozilla has the src, you can have m:/mozilla/gcc322 and p:/mozilla/gcc335, tvlink all three to y:, then run ../configure from either the gcc322 or gcc335 subdirectory on y:, or any other used to differentiate builds, and the built output will end up on the linked drive. Perl, however, differs in that the default is to build in the source directory. To build elsewhere, you need a working ln -s and have to run Configure -Dmksymlinks from the directory you want to build in. It parses MANIFEST in the Perl root directory to create a symlinked build tree and put all the output there. With TVFS, if you tvlink the perl-5.8.5 on c: with -dr (and any other options except -w to make the source tree read-only), and the one on d: with -drw, building on x: will fail if the process cd's into a subdirectory that doesn't exist on d: and tries to write to the existing one on c:. But if you tvlink the perl on d: with -cdrw, it should create the missing parallel tree on d: as it goes along. If that fails because of some quirk in Perl's build system, a script that parses MANIFEST for the directory structure and creates it on d: before running Configure might be the simplest workaround. h~ **= Email 25 ==========================** Date: Sat, 27 Nov 2004 22:03:49 +0100 From: Stefan.Neis at t-online.de Subject: Re: Does link do anything Hi, > link is hardlinks, i.e. not supported. The function is a stub in LIBC > and will just do the parameter validation before returning with errno > set to ENOSYS. Wouldn't it generally be more helpful to just call symlink instead of giving an error message? AFAIK, the difference boils down to hardlinks never being left pointing to nowhere if the "original" gets deleted and I admit I'm more happy with applications breaking in that case than applications not running at all because they can't make hardlinks. Regards, stefan **= Email 26 ==========================** Date: Sat, 27 Nov 2004 21:57:52 +0100 From: Stefan.Neis at t-online.de Subject: Re: BSD Makefiles Hi, > What do I need to be able to use BSD Makefiles? The corresponding make utility. It should be automatically compiled if you compile e.g. Posix/2 libraries. > > I have a Makefile for SYSLOG which consists of:- (snipp) > ..include > > What is this bsd.prog.mk ? Another makefile being included by the current one, probably taken from some "standard" directory for makefile fragments. It's probably hard to make use of those files without a BSD system next to you where you can search for such files and copy them as needed... Regards, Stefan **= Email 27 ==========================** Date: Sun, 28 Nov 2004 09:52:21 +0100 (CET) From: "Yuri Dario" Subject: Re: Does link do anything Hi Knut, >Can you explain how to reproduce this and I'll fix it. :-) this is a simple testcase: #include #include void main(void) { int rc; char buff[260]; FILE* d = fopen( "dummy.txt", "w"); fprintf( d, "some text\n"); fclose(d); unlink( "test\\dummy.txt"); rc = symlink( "dummy.txt", "test\\dummy.txt"); printf( "symlink rc=%d\n", rc); rc = realpath("dummy.txt",buff); printf( "realpath rc=%d\n", rc); rc = realpath("test\\dummy.txt",buff); printf( "realpath rc=%d\n", rc); } the problem is in the SYMLINK EA: E:\rd\MySQL\test\symlink>eatool -l test\dummy.txt test\dummy.txt: SYMLINK: "dummy.txt" the linked file points to a relative path: since the current directory can change, I suggest to use full paths in symlinked files. Bye, Yuri Dario /* * member of TeamOS/2 - Italy * http://www.os2power.com/yuri * http://www.teamos2.it */ **= Email 28 ==========================** Date: Sun, 28 Nov 2004 11:47:26 +1100 (EST) From: Andrew MacIntyre Subject: Re: Building Python 2.3.4 with Innotek GCC 3.3.5 beta 1 On Sat, 27 Nov 2004, Paul Smedley wrote: > A few things required change: > 1) in pc/os2emx - pyconfig.h needs editing and the '#define intptr_t' > removing as it's already defined in Innotek GCC That can be dealt with fairly easily. > 2) in python - thread.c - I had to comment out: > #ifdef _POSIX_THREADS > /*#include "thread_pthread.h" innote */ > #endif > as it appears that Innotek GCC defines that is has _POSIX_THREADS and > this code gets executed as well as the _OS2_THREADS when building thread.c Interesting. I can probably deal with this without too much drama. > 3) in pc/os2emx - I had to comment out > $(EXEOPT) -aq $(PYTHON.EXE) -h$(NFILES) > & $(EXEOPT) -aq $(PYTHONPM.EXE) -h$(NFILES) > Innotek GCC's emxbind doesn't support -a - and with just -q I got an > error 'emxbind: The input and output files have the same name' Without knowing what the Innotek libc's default number of file handles is its hard to assess the impact of this. At least one of the python regression tests needs 100 file handles. > It seems to run OK but running 'make test' I get > <...snip...> > test_fnmatch > test_fork1 > > !!!deadlock!!! LIBC Heap Recursive mutex! > > !!!deadlock!!! LIBC Heap Recursive mutex! > make: *** [test] Error 127 > > not sure if this is due to the handles not being set correctly in > python.exe or not.... I doubt it - I strongly suspect a libc bug/limitation. Hopefully Knut can investigate. I'd be interested in knowing how you get on with building Python 2.4c1. I know of a couple of things that need doing (I haven't had much time to track the CVS changes just lately), but they shouldn't affect the basic build. ------------------------------------------------------------------------- Andrew I MacIntyre "These thoughts are mine alone..." E-mail: andymac at bullseye.apana.org.au (pref) | Snail: PO Box 370 andymac at pcug.org.au (alt) | Belconnen ACT 2616 Web: http://www.andymac.org/ | Australia **= Email 29 ==========================** Date: Sun, 28 Nov 2004 12:48:39 +1100 (EST) From: Andrew MacIntyre Subject: Re: System V Semaphore emulation On Tue, 23 Nov 2004, Knut St. Osmundsen wrote: > Andrew MacIntyre wrote: > > The big trap with these, as far as I understand, is that they are > > persistent - ie they survive after all referencing processes die, and can > > be picked up later by new processes. This applies to the Sys V shared > > memory (shm*()) and message queues as well. > Interesting, didn't know about that aspect of the ipc stuff. > > OTOH I'm not sure if SuS/Posix specifies it that way. It's quite > possible that they leave this behaviour open. I don't know about the official specs, but I encountered this behaviour on Solaris (which is Sys5R4). There's an ipcrm for dealing with unwanted leftovers, and I had to use it in a cron job to deal with one particular set of circumstances. FreeBSD has an ipcrm too. VV's implementation doesn't, at least as distributed with his PostgreSQL port, but the daemon can be restarted. > In any case I don't wanna > have a wine solution with a libc daemon if it can be helped, nor do I > wanna have resources which aren't automatically cleaned up when we exit > a process. If you decide to support the full semantics, my approach would be to have a separate daemon (a la VV's implementation), which must be explicitly started by the user. If the API functions can't contact the daemon, they should report failure - which is what happens on FreeBSD if the kernel modules aren't compiled in. AIRC, Solaris has some sort of autoloading of most kernel modules. An implementation strategy that I had thought about is to use a single instance DLL (ie the DLL has its own data segment) which stays alive as long as some libc linked process is running. This would provide semantics suitable for the likes of PostgreSQL. A simple extention is then to add the "daemon" to keep it alive for those circumstances that need it. > The [shm|msg|sem]ctl(,IPC_RMID,) calls can be implemented although OS/2 > doesn't support removing any shared resource from all processes > (officially anyway). This might mean that the semaphore implementation > won't be very fast for instance. Have the IPC DLL "forget" about the resources, and document the fact that processes that have obtained a resource can't be divested of it. Invite anyone discovering a problem to present a test case... In my somewhat limited experience, using the IPC_RMID operation is used by a cooperative resource manager (eg PostgreSQL's postmaster) after its related processes have been thought to have terminated, or by an administrator trying to clean up the mess of a crash. In the absence of evidence to the contrary, I can't see much of a use case for fully supporting the forcible removal of resources. ------------------------------------------------------------------------- Andrew I MacIntyre "These thoughts are mine alone..." E-mail: andymac at bullseye.apana.org.au (pref) | Snail: PO Box 370 andymac at pcug.org.au (alt) | Belconnen ACT 2616 Web: http://www.andymac.org/ | Australia **= Email 30 ==========================** Date: Sun, 28 Nov 2004 11:55:58 +1100 (EST) From: Andrew MacIntyre Subject: Re: Building Python 2.3.4 with Innotek GCC 3.3.5 beta 1 On Sat, 27 Nov 2004, John Poltorak wrote: > I wonder if we are nearing a stage where we can build python from the > original source using autoconf, configure and make without requiring OS/2 > specific makefiles... Quite possibly, but the import machinery for coping with the 8.3 restrictions applying to DLLs (used for dynamically loadable extension modules) is currently dependant on definitions in the custom configuration file (PC/os2emx/pyconfig.h). ------------------------------------------------------------------------- Andrew I MacIntyre "These thoughts are mine alone..." E-mail: andymac at bullseye.apana.org.au (pref) | Snail: PO Box 370 andymac at pcug.org.au (alt) | Belconnen ACT 2616 Web: http://www.andymac.org/ | Australia **= Email 31 ==========================** Date: Sun, 28 Nov 2004 10:58:34 +0100 (CET) From: "Yuri Dario" Subject: Re: Does link do anything Hi, >the linked file points to a relative path: since the current directory can change, I suggest to use >full paths in symlinked files. I'm reading something more about hard and sym links, I now understand why absolute paths are unusable: since moving directories must not break symlinks, paths must be relative. Maybe in the testcase, test\dummy.txt should symlink ..\dummy.txt (instead of dummy.txt as is now). Bye, Yuri Dario /* * member of TeamOS/2 - Italy * http://www.os2power.com/yuri * http://www.teamos2.it */ **= Email 32 ==========================** Date: Sun, 28 Nov 2004 21:23:06 +1030 From: Paul Smedley Subject: Re: Building Python 2.3.4 with Innotek GCC 3.3.5 beta 1 Hi Andrew, Andrew MacIntyre wrote: > I'd be interested in knowing how you get on with building Python 2.4c1. > I know of a couple of things that need doing (I haven't had much time to > track the CVS changes just lately), but they shouldn't affect the basic > build. python.exe, pgen.exe, pythonpm.exe & python24.dll all get successfully built using the same changes as for python 2.3.4 However, appears I got carried away with the completion of compilation - noticed with 2.4rc1 the following: Creating .DEF file: out/optimize/fpectl_m.def gcc -Zmt -Zcrtdll -L. -lgcc -s -Zomf -Zdll -o fpectl.pyd out/optimize/fpectlmodu le.obj out/optimize/fpectl_m.def python24.lib -lsocket weakld: error: Unresolved symbol (UNDEF EXPORT) 'initfpectl'. weakld: info: The symbol is referenced by: E:/DEV/Python-2.4c1/PC/os2emx/out/optimize/fpectl_m.def Ignoring unresolved externals reported from weak prelinker. ILink : error LNK2022: initfpectl (alias initfpectl) : export undefined There was 1 error detected make[1]: *** [fpectl.pyd] Error 1 rm out/optimize/fpectlmodule.obj out/optimize/fpectl_m.def make[1]: Leaving directory `/dev/python-2.4c1/pc/os2emx' make: *** [python_noncore] Error 2 Which rang a bell as happening when I build 2.3.4 as well I just failed to report it in my last post :( Any ideas? **= Email 33 ==========================** Date: Sun, 28 Nov 2004 11:06:54 +0000 From: John Poltorak Subject: Re: Building Python 2.3.4 with Innotek GCC 3.3.5 beta 1 On Sat, Nov 27, 2004 at 04:11:26PM +1030, Paul Smedley wrote: > Hi All, > I just got Python 2.3.4 to build with Knut's latest GCC. > > A few things required change: > 1) in pc/os2emx - pyconfig.h needs editing and the '#define intptr_t' > removing as it's already defined in Innotek GCC > 2) in python - thread.c - I had to comment out: > #ifdef _POSIX_THREADS > /*#include "thread_pthread.h" innote */ > #endif > as it appears that Innotek GCC defines that is has _POSIX_THREADS and > this code gets executed as well as the _OS2_THREADS when building thread.c > 3) in pc/os2emx - I had to comment out > $(EXEOPT) -aq $(PYTHON.EXE) -h$(NFILES) > & $(EXEOPT) -aq $(PYTHONPM.EXE) -h$(NFILES) Have you forgotten to mention something? I take it the last change was to the makefile.. I made those three changes but still ended with this error:- gcc -Zmt -Wall -I. -I../../Include -s -O3 -fomit-frame-pointer -mprobe -DPy_BUILD_CORE -DNDEBUG -Zomf -c ../../Objects/longobject.c -o out/optimize/longobject.obj .../../Objects/longobject.c: In function `long_format': .../../Objects/longobject.c:1141: warning: dereferencing type-punned pointer will break strict-aliasing rules .../../Objects/longobject.c: At top level: .../../Objects/longobject.c:2702: error: `long_int' redeclared as different kind of symbol g:/usr/include/sys/types.h:351: error: previous declaration of `long_int' make: *** [out/optimize/longobject.obj] Error 1 Any suggestions about what else to change? > Cheers, > > Paul. -- John **= Email 34 ==========================** Date: Sun, 28 Nov 2004 11:09:21 +0000 From: John Poltorak Subject: [madodel at ptdprolog.net: [VOICENWS] SW: MySQL 4.0.18 for OS/2] --F0BwYIV0LwSY2Fe/ Content-Type: text/plain; charset=us-ascii Just thought some people here may be interested in news about a new MySQL release... -- John --F0BwYIV0LwSY2Fe/ Content-Type: message/rfc822 Received: from voice2.os2voice.org by mail.warpix.org (IBM OS/2 SENDMAIL VERSION 2.03/2.0) id GAA374.25; Sun, 28 Nov 2004 06:20:36 GMT From: madodel at ptdprolog.net Received: from smtp16.mailnet.ptd.net (smtp16.mailnet.ptd.net [204.186.29.136]) by os2voice.org (Weasel v1.71) for ; 27 Nov 2004 21:24:23 Received: (qmail 26298 invoked by uid 50005); 28 Nov 2004 02:24:19 -0000 Received: from madodel at ptdprolog.net by smtp16.mailnet.ptd.net by uid 50002 with qmail-scanner-1.20 (uvscan: v4.2.40/v4409. Clear:RC:0(24.238.84.87):. Processed in 0.467036 secs); 28 Nov 2004 02:24:19 -0000 Received: from 24.238.84.87.res-cmts.sth.ptd.net (HELO MEDION) (authenticated:madodel at [24.238.84.87]) (envelope-sender ) by smtp16.mailnet.ptd.net (qmail-ldap-1.03) with SMTP for ; 28 Nov 2004 02:24:18 -0000 Reply-to: mc6530 at mclink.it, feedback at os2voice.org Date: Sat, 27 Nov 2004 21:24:17 -0500 To: news at os2voice.org Subject: [VOICENWS] SW: MySQL 4.0.18 for OS/2 X-Mailer: MR/2 Internet Cruiser Edition for OS/2 v2.60b X-Qmail-Scanner-Message-ID: <110160865865326281 at smtp16.mailnet.ptd.net> Message-ID: <0077063600.000071M1 at os2voice.org> From: mc6530DESPAM at mclink.it MySQL 4.0.18 for OS/2 is now available for public download: http://www.os2power.com/yuri/ The MySQL database server is the world's most widely used open source database. Its ingenious software architecture makes it extremely fast and easy to customize. MySQL is available for free under the GNU General Public Licence (GPL). MySQL is provided by MySQL AB. NOTE this code has been previously released for private test to mysql-os2 members: if you are already running that code, you don't need to upgrade. If you upgrade from 3.x, please read the manual for table updates. New features ------------ Named pipes. Data replication. Binaries can be moved everywhere. InnoDB database engine is included. Enabled usage of high memory arena for memory allocation. Binaries are compiled for Pentium processor (or higher), code is tuned for Pentium III processors (or higher). File support for 64bit access (>2GB databases). System requirements: -------------------- - OS/2 Warp 3 with FP ?? or later, OS/2 Warp 4 with FP ?? or later, OS/2 Warp Server for e-Business, eComStation 1.0,1.1 - TCPIP 4.x installed (requires 32-bit tcpip stack) - WarpIN installer 1.0.2 (ftp://ftp.netlabs.org/pub/warpin/warpin-1-0-2.exe) - Pentium (or higher) processor is required (no 386/486) Support ------- Please subscribe to mysql-os2 at www.yahoogroups.com to get in contact with other users using this port. http://www.yahoogroups.com/group/mysql-os2 Donations --------- Since this software is ported for free, donations are welcome! you can use PayPal to donate me and support OS/2 developement. You can get also an extended support, which is not free and subject to custom rates. Url: http://www.os2power.com/yuri/ From: Yuri Dario (mc6530DESPAM at mclink.it) Submitted with the form on http://www.os2world.com/ -- Warpstock Europe 2004, Arnhem, The Netherlands, November 26-28th, 2004 http://www.warpstock.net Warpstock 2005 - Where/When? Stay tuned to http://www.warpstock.org [Moderator's note: All posts are sent without guarantee to the accuracy of the content. We try to verify details and URLs but this is an entirely volunteer run list, so 100% fact checking and the quality/useability of products announced here is impossible. If you respond to this post please remove the DESPAM from the poster's email addresses. Please do not send requests for information about a specific post to the moderator unless it is an update or I sent it.] -- To unsubscribe yourself from this list, send the following message to majormajor at os2voice.org unsubscribe news end Or, visit http://www.os2voice.org/MailingLists.html If you have an announcement you would like posted to the VOICE News list, please send it to submit at os2voice.org. Please include a valid reply address and a real contact name. If you wish to comment on this post, please reply to feedback at os2voice.org --F0BwYIV0LwSY2Fe/-- **= Email 35 ==========================** Date: Sun, 28 Nov 2004 12:24:47 +0100 From: Stefan.Neis at t-online.de Subject: Re: Does link do anything Hi, > I'm reading something more about hard and sym links, I now understand why absolute paths are > unusable: since moving directories must not break symlinks, paths must be relative. > Maybe in the testcase, test\dummy.txt should symlink ..\dummy.txt (instead of dummy.txt as > is now). That's something often used on Unix systems: The file a link links to is given with a path relative to the links location (i.e. ..\dummy.txt in your case), so that's certainly a good idea. Regards, Stefan **= Email 36 ==========================** Date: Sun, 28 Nov 2004 11:38:01 +0000 From: John Poltorak Subject: Building Zope with GCC 3.3.5 Just thought I'd try building Zope with gcc 3.3.5... First time I got:- g:/usr/bin/gcc.exe -Zomf -Zmt -O3 -fomit-frame-pointer -mprobe -Wall -IG:/ux2bs/workdir/Zope-2.7.3-0/lib/Components/ExtensionClass/src -I/usr/local/python/Include -c AccessControl/cAccessControl.c -o G:/ux2bs/workdir/Zope-2.7.3-0/build-base/python-2.3/build-temp/accesscontrol/caccesscontrol.obj In file included from /usr/local/python/Include/Python.h:36, from G:/ux2bs/workdir/Zope-2.7.3-0/lib/Components/ExtensionClass/src/ExtensionClass.h:94, from AccessControl/cAccessControl.c:51: g:/usr/include/stdlib.h:324: error: conflicting types for `intptr_t' /usr/local/python/Include/pyconfig.h:86: error: previous declaration of `intptr_t' command 'gcc' failed with exit status 1 error: command 'gcc' failed with exit status 1 make: *** [build] Error 1 This has already appeared as a problem building Python so the declaration of intptr_t in Python.h was commented out. Next attempt:- running build_ext building 'AccessControl.cAccessControl' extension creating G:/ux2bs/workdir/Zope-2.7.3-0/build-base/python-2.3/build-temp creating G:/ux2bs/workdir/Zope-2.7.3-0/build-base/python-2.3/build-temp/accesscontrol g:/usr/bin/gcc.exe -Zomf -Zmt -O3 -fomit-frame-pointer -mprobe -Wall -IG:/ux2bs/workdir/Zope-2.7.3-0/lib/Components/ExtensionClass/src -I/usr/local/python/Include -c AccessControl/cAccessControl.c -o G:/ux2bs/workdir/Zope-2.7.3-0/build-base/python-2.3/build-temp/accesscontrol/caccesscontrol.obj writing G:/ux2bs/workdir/Zope-2.7.3-0/build-base/python-2.3/build-temp/accesscontrol/cAccessC.def g:/usr/bin/gcc.exe -Zomf -Zmt -Zcrtdll -Zdll -s G:/ux2bs/workdir/Zope-2.7.3-0/build-base/python-2.3/build-temp/accesscontrol/caccesscontrol.obj G:/ux2bs/workdir/Zope-2.7.3-0/build-base/python-2.3/build-temp/accesscontrol/cAccessC.def -L/usr/local/python/Config -lpython23 -lgcc -o G:/ux2bs/workdir/Zope-2.7.3-0/build-base/python-2.3/build-lib/AccessControl/cAccessC.pyd weakld: error: Unresolved symbol (UNDEF) '_PyErr_Occurred'. weakld: info: The symbol is referenced by: G:\ux2bs\workdir\Zope-2.7.3-0\build-base\python-2.3\build-temp\accesscontrol\caccesscontrol.obj weakld: error: Unresolved symbol (UNDEF) '_PyExc_AttributeError'. weakld: info: The symbol is referenced by: G:\ux2bs\workdir\Zope-2.7.3-0\build-base\python-2.3\build-temp\accesscontrol\caccesscontrol.obj A large number of similar errors occurred the build stpped with:- weakld: info: The symbol is referenced by: G:\ux2bs\workdir\Zope-2.7.3-0\build-base\python-2.3\build-temp\accesscontrol\caccesscontrol.obj Ignoring unresolved externals reported from weak prelinker. G:\ux2bs\workdir\Zope-2.7.3-0\build-base\python-2.3\build-temp\accesscontrol\caccesscontrol.obj(caccesscontrol.obj) : error LNK2029: "_PySequence_GetItem" : unresolved external G:\ux2bs\workdir\Zope-2.7.3-0\build-base\python-2.3\build-temp\accesscontrol\caccesscontrol.obj(caccesscontrol.obj) : error LNK2029: "_PySequence_Contains" : unresolved external G:\ux2bs\workdir\Zope-2.7.3-0\build-base\python-2.3\build-temp\accesscontrol\caccesscontrol.obj(caccesscontrol.obj) : error LNK2029: "_PyString_FromString" : unresolved external G:\ux2bs\workdir\Zope-2.7.3-0\build-base\python-2.3\build-temp\accesscontrol\caccesscontrol.obj(caccesscontrol.obj) : error LNK2029: "_PyCObject_AsVoidPtr" : unresolved external