Date: Tue, 11 Jan 2005 00:04:20 EST-10EDT,10,-1,0,7200,3,-1,0,7200,3600 Subject: [UnixOS2_Archive] No. 498 ************************************************** Monday 10 January 2005 Number 498 ************************************************** Subjects for today 1 Re: Webmin and CPAN : Lyn St George" 2 Re: emx\include\netdb.h:123: parse error before `int' : Stefan.Neis at t-online.de 3 Re: Webmin and CPAN : John Poltorak 4 Re: WINDOWCOMPAT : Stefan.Neis at t-online.de 5 Re: Webmin and CPAN : Lyn St George" 6 Re: Webmin and CPAN : John Poltorak 7 Re: Python question : Stefan.Neis at t-online.de 8 Re: emx\include\netdb.h:123: parse error before `int' : John Poltorak 9 Re: Lowercase variable names : Andreas Buening 10 Re: Problem running Autoconf : Andreas Buening 11 Re: Webmin & FHS : Andreas Buening 12 Re: Openssl-0.9.7e : Stefan.Neis at t-online.de 13 Re: Perl & CPAN.pm : Dave Yeo" 14 Re: emx\include\netdb.h:123: parse error before `int' : Dave Yeo" 15 Re: Re: Openssl-0.9.7e : Dave Yeo" 16 Re: emx\include\netdb.h:123: parse error before `int' : Dave Yeo" 17 Re: Perl & CPAN.pm : John Poltorak 18 Re: Webmin and CPAN : Dave Yeo" 19 Re: emx\include\netdb.h:123: parse error before `int' : John Poltorak 20 Re: Re: Openssl-0.9.7e : John Poltorak 21 Tracking source changes - was - Re: emx\include\netdb.h:123: parse error before `int' : lsunley at mb.sympatico.ca 22 Re: Re: Openssl-0.9.7e : Stefan.Neis at t-online.de 23 Re: emx\include\netdb.h:123: parse error before `int' : Stefan.Neis at t-online.de 24 Re: Re: Openssl-0.9.7e : Dave Yeo 25 Re: Tracking source changes - was - Re: : Dave Yeo 26 Re: Re: Openssl-0.9.7e : Dave Yeo 27 Re: Tracking source changes - was - Re: emx\include\netdb.h:123: parse error before `int' : lsunley at mb.sympatico.ca 28 Re: Webmin and CPAN : Lyn St George" 29 Re: Tracking source changes - was - Re: emx\include\netdb.h:123: parse error before `int' : John Poltorak 30 Re: Re: Openssl-0.9.7e : John Poltorak 31 Re: Webmin and CPAN : John Poltorak 32 Re: distcc : John Poltorak 33 Re: Re: Openssl-0.9.7e : Stefan.Neis at t-online.de 34 Re: Re: Openssl-0.9.7e : John Poltorak 35 Re: Tracking source changes - was - Re: emx\include\netdb.h:123: parse error before `int' : Stefan.Neis at t-online.de 36 Re: Lowercase variable names : John Poltorak 37 Re: Re: Openssl-0.9.7e : Brian Havard" **= Email 1 ==========================** Date: Sun, 09 Jan 2005 13:20:12 +0000 (GMT) From: "Lyn St George" Subject: Re: Webmin and CPAN On Sun, 9 Jan 2005 10:37:11 +0000, John Poltorak wrote: >> I've installed the latest 1.170 and found that things are different, again. > >I installed 1.170 using your patches for 1.140! > >> This problem is in web-lib.pl and is fixed by adding at line 145: >> >> if ($^O =~ /os2/i) { >> unless ( -d $tmp_base ) { >> mkdir($tmp_base, 0755) >> } >> } >> else { >> >> plus another right curly brace at the end of that subroutine. >> However, now it just sits and spins its wheels ... > >So is it really fixed? Well, it plainly needs some more investigation and work. I'll look around at what needs fixing and then send you a new patch. In the meantime, you could see what you think of this version running on my local OS/2 box, at https://merlin.zolotek.net:10055, login ux2/ux2 (hopefully it's been sufficiently defanged - I don't know who is lurking on this list ...) This uses my Zedpanel theme (as used for my hosting customers), and will need to have the modules sorted out for relevance on OS/2. - Lyn **= Email 2 ==========================** Date: Sun, 9 Jan 2005 15:06:17 +0100 From: Stefan.Neis at t-online.de Subject: Re: emx\include\netdb.h:123: parse error before `int' Hi, > Part of host.c:- > > #include > (*) > #ifndef WINDOWS > #include <======================= line 33 > #endif > > #include > #include > #ifdef HAVE_STRING_H > # include > #else > # include > #endif > #include > #include (**) > > I thought WGET was buildable on OS/2 straight out of the box... Apparently not the latest(?) version. Move that line which I marked with (**) closer to the beginning of the file (e.g. to the line I marked with (*)) and everything should be fine. Regards, Stefan **= Email 3 ==========================** Date: Sun, 9 Jan 2005 14:06:35 +0000 From: John Poltorak Subject: Re: Webmin and CPAN On Sun, Jan 09, 2005 at 01:20:12PM +0000, Lyn St George wrote: > On Sun, 9 Jan 2005 10:37:11 +0000, John Poltorak wrote: > > >> I've installed the latest 1.170 and found that things are different, again. > > > >I installed 1.170 using your patches for 1.140! > > > >> This problem is in web-lib.pl and is fixed by adding at line 145: > >> > >> if ($^O =~ /os2/i) { > >> unless ( -d $tmp_base ) { > >> mkdir($tmp_base, 0755) > >> } > >> } > >> else { > >> > >> plus another right curly brace at the end of that subroutine. > >> However, now it just sits and spins its wheels ... > > > >So is it really fixed? > > Well, it plainly needs some more investigation and work. > > I'll look around at what needs fixing and then send you a new > patch. Thanks, I'd appreciate that, I getting hungry for what Webmin should be able to provide... > In the meantime, you could see what you think of this version > running on my local OS/2 box, at > https://merlin.zolotek.net:10055, login ux2/ux2 (hopefully it's been > sufficiently defanged - I don't know who is lurking on this list ...) That's very impressive! And all done through Webmin?... > This uses my Zedpanel theme (as used for my hosting customers), > and will need to have the modules sorted out for relevance on OS/2. It would be nice to come up with something which resembles a Warp desktop. > - > Lyn > -- John **= Email 4 ==========================** Date: Sun, 9 Jan 2005 15:12:43 +0100 From: Stefan.Neis at t-online.de Subject: Re: WINDOWCOMPAT Hi, > On Fri, Jan 07, 2005 at 09:57:56PM +0000, Dave Saville wrote: > > On Fri, 7 Jan 2005 13:40:05 +0000, John Poltorak wrote: > > > > > > > >Is there a way of specifying WINDOWCOMPAT via the environment so that you > > >don't get warning msgs from LINK386 when application type is not > > >mentioned? > > > > Should one not be setting this in a .def file? > > I'd like to do this globally for every build... Maybe I can I create a > unixos2.def, but then how do I get the build to recognise it? > Setting CXXFLAGS?... No it needs to be linked into the executable just like any object file. I guess, Henry's(?) suggestion about using "-Zlinker /PMTYPE:VIO"(which could be added to LDFLAGS gobally) is easier to integrate globally. Regards, Stefan **= Email 5 ==========================** Date: Sun, 09 Jan 2005 14:19:43 +0000 (GMT) From: "Lyn St George" Subject: Re: Webmin and CPAN On Sun, 9 Jan 2005 10:37:11 +0000, John Poltorak wrote: >> This problem is in web-lib.pl and is fixed by adding at line 145: >> >> if ($^O =~ /os2/i) { >> unless ( -d $tmp_base ) { >> mkdir($tmp_base, 0755) >> } >> } >> else { >> >> plus another right curly brace at the end of that subroutine. >> However, now it just sits and spins its wheels ... > >So is it really fixed? No. That patch was produced at 3 in the morning, and missed the main point. The cpan module used to work in earlier versions without any problems or patching, though this tempname routine is only different in that it checks the number of attempts and fails after 10. I'll try to sort this out today, if possible. - Lyn **= Email 6 ==========================** Date: Sun, 9 Jan 2005 14:45:53 +0000 From: John Poltorak Subject: Re: Webmin and CPAN On Sun, Jan 09, 2005 at 02:19:43PM +0000, Lyn St George wrote: > On Sun, 9 Jan 2005 10:37:11 +0000, John Poltorak wrote: > > >> This problem is in web-lib.pl and is fixed by adding at line 145: > >> > >> if ($^O =~ /os2/i) { > >> unless ( -d $tmp_base ) { > >> mkdir($tmp_base, 0755) > >> } > >> } > >> else { > >> > >> plus another right curly brace at the end of that subroutine. > >> However, now it just sits and spins its wheels ... > > > >So is it really fixed? > > No. That patch was produced at 3 in the morning, and missed the > main point. The cpan module used to work in earlier versions without > any problems or patching, though this tempname routine is only > different in that it checks the number of attempts and fails after 10. > I'll try to sort this out today, if possible. I'll revert back to an older version of Webmin if that works. I'm more interested in having an older version of a working module than the latest which needs fixing... > > - > Lyn > -- John **= Email 7 ==========================** Date: Sun, 9 Jan 2005 16:04:11 +0100 From: Stefan.Neis at t-online.de Subject: Re: Python question Hi, > > ERROR: > > Traceback (most recent call last): > > File "BitTorrent/RawServer.py", line 210, in listen_forever > > events = self.poll.poll(period * timemult) > > File "BitTorrent/selectpoll.py", line 43, in poll > > r, w, e = select(self.rlist, self.wlist, [], timeout) > > ValueError: filedescriptor out of range in select() > > I thought FD_SETSIZE was being set to 512, but need to check. Well, I get that error for handle values in the range between 257 and 512.. **= Email 8 ==========================** Date: Sun, 9 Jan 2005 15:30:42 +0000 From: John Poltorak Subject: Re: emx\include\netdb.h:123: parse error before `int' On Sun, Jan 09, 2005 at 03:06:17PM +0100, Stefan.Neis at t-online.de wrote: > Hi, > > > Part of host.c:- > > > > #include > > (*) > > #ifndef WINDOWS > > #include <======================= line 33 > > #endif > > > > #include > > #include > > #ifdef HAVE_STRING_H > > # include > > #else > > # include > > #endif > > #include > > #include (**) > > > > I thought WGET was buildable on OS/2 straight out of the box... > > Apparently not the latest(?) version. > Move that line which I marked with (**) closer to the beginning > of the file (e.g. to the line I marked with (*)) and everything > should be fine. It resolves that error but leads to this one further along:- /bin/sh ../libtool --mode=link gcc -O2 -Wall -Wno-implicit -o wget cmpt.o connect.o convert.o cookies.o ftp.o ftp-basic.o ftp-ls.o ftp-opie.o getopt.o hash.o headers.o host.o html-parse.o html-url.o http.o init.o log.o main.o gen-md5.o gnu-md5.o netrc.o progress.o rbuf.o recur.o res.o retr.o safe-ctype.o snprintf.o url.o utils.o version.o -lintl -lsocket mkdir .libs gcc -O2 -Wall -Wno-implicit -o wget cmpt.o connect.o convert.o cookies.o ftp.o ftp-basic.o ftp-ls.o ftp-opie.o getopt.o hash.o headers.o host.o html-parse.o html-url.o http.o init.o log.o main.o gen-md5.o gnu-md5.o netrc.o progress.o rbuf.o recur.o res.o retr.o safe-ctype.o snprintf.o url.o utils.o version.o c:/usr/lib/intl.a -lsocket -Lc:/usr/lib -Lc:/usr/lib utils.o: Undefined symbol _setsid referenced from text segment make[1]: *** [wget] Error 1 That is without Posix/2. If I include it, I get:- /bin/sh ../libtool --mode=link gcc -O2 -Wall -Wno-implicit -o wget cmpt.o connect.o convert.o cookies.o ftp.o ftp-basic.o ftp-ls.o ftp-opie.o hash.o headers.o host.o html-parse.o html-url.o http.o init.o log.o main.o gen-md5.o gnu-md5.o netrc.o progress.o rbuf.o recur.o res.o retr.o safe-ctype.o snprintf.o url.o utils.o version.o -lintl mkdir .libs gcc -O2 -Wall -Wno-implicit -o wget cmpt.o connect.o convert.o cookies.o ftp.o ftp-basic.o ftp-ls.o ftp-opie.o hash.o headers.o host.o html-parse.o html-url.o http.o init.o log.o main.o gen-md5.o gnu-md5.o netrc.o progress.o rbuf.o recur.o res.o retr.o safe-ctype.o snprintf.o url.o utils.o version.o c:/usr/lib/intl.a -Lc:/usr/lib -Lc:/usr/lib utils.o: Undefined symbol _setsid referenced from text segment c:\ux2bs\posix2\lib/cExt.a(itimer.o): Undefined symbol __beginthread referenced from text segment make[1]: *** [wget] Error 1 > Regards, > Stefan > > -- John **= Email 9 ==========================** Date: Sun, 09 Jan 2005 15:35:22 +0100 From: Andreas Buening Subject: Re: Lowercase variable names John Poltorak wrote: > > How would I set variable names to lowercase in a cmd file? Not at all. > AIUI shell scripts can do this so maybe I can achieve this by running a > shell script using EXTPROC. This can never work because any other shell would be executed as a child process. > Or is there an alternative way of creating lowercase variable names? The only way is to start a Unix shell, to set the lower case variables and finally to start another cmd within that shell. Bye, Andreas **= Email 10 ==========================** Date: Sun, 09 Jan 2005 15:35:46 +0100 From: Andreas Buening Subject: Re: Problem running Autoconf John Poltorak wrote: > > I got this error whilst trying to build Apache2 (v2.0.52) :- > > httpd-2.0.52 > Using:- autoconf (GNU Autoconf) 2.59 (release 2 for OS/2) > SYS1041: The name . is not recognized as an > internal or external command, operable program or batch file. > > How can I find out what is causing this? It's nearly impossible to help you without further information. At least, the fact that cmd is called seems to be an error. Bye, Andreas **= Email 11 ==========================** Date: Sun, 09 Jan 2005 15:36:00 +0100 From: Andreas Buening Subject: Re: Webmin & FHS John Poltorak wrote: > > Installing Webmin involves the use of three directories, one for the > program, one for the config files and one for the logs. I'm none too sure > how FHS compliant the default installation is - it uses /etc/webmin for > config files and /var/webmin for logs. The directory from which webmin is > installed is the default application directory. > > What would be more suitable? How about these:- ? /etc/webmin is okay, and /var/log/webmin for logfiles. The philosphy of FHS doesn't provide application directories for a single application. Binaries should go to /usr/bin (or /usr/local/bin), Manpages to /usr/share/man, other files to /usr/share/webmin. Bye, Andreas **= Email 12 ==========================** Date: Sun, 9 Jan 2005 19:05:17 +0100 From: Stefan.Neis at t-online.de Subject: Re: Openssl-0.9.7e Hi, I just did a static build of Openssl-0.9.7e which worked OOTB. Those problems with str(n)casecmp only occur when trying to build DLLs, I guess. To me, it looks like o_str.c (and the corresponding o_str.o) define two symbols which are only ever used on old VMS versions (which don't have anything like str(n)casecmp), so the easy fix seems to be to just completely remove that file from the list of files to be compiled/linked on OS/2. Or maybe the long term plan of OpenSSL maintainers is to completely get rid of all those str(n)casecmp defines and always call OPENSSL_strncasecmp on all platforms, then o_str.c should be fixed to do the right thing on OS/2. Regards, Stefan **= Email 13 ==========================** Date: Sun, 09 Jan 2005 11:16:30 -0800 From: "Dave Yeo" Subject: Re: Perl & CPAN.pm On Sun, 9 Jan 2005 12:00:41 +0000, John Poltorak wrote: >Does anyone else have any problems with using the CPAN module with Perl? I've used the CPAN module (though not for a year or more) and never had any problems with corrupt files. Did have some other problems eg make test failing which I had to cleanup by hand when installing some modules but otherwise it worked pretty well. This was with my own build of perl Dave **= Email 14 ==========================** Date: Sun, 09 Jan 2005 11:22:29 -0800 From: "Dave Yeo" Subject: Re: emx\include\netdb.h:123: parse error before `int' On Sun, 9 Jan 2005 15:30:42 +0000, John Poltorak wrote: >utils.o: Undefined symbol _setsid referenced from text segment >make[1]: *** [wget] Error 1 I hate that _setsid error. I've had it come and go just by changing optimization, I really think it is a bug in EMX. I usually end up just commenting out the setsid line in the source. setsid doesn't do anything under OS/2 anyways as far as I know Dave **= Email 15 ==========================** Date: Sun, 09 Jan 2005 11:24:15 -0800 From: "Dave Yeo" Subject: Re: Re: Openssl-0.9.7e On Sun, 9 Jan 2005 19:05:17 +0100, Stefan.Neis at t-online.de wrote: > >Or maybe the long term plan of OpenSSL maintainers is to >completely get rid of all those str(n)casecmp defines and always >call OPENSSL_strncasecmp on all platforms, then o_str.c should >be fixed to do the right thing on OS/2. I got the feeling that this is there long term plan and another ifelse or two should be added to o_str.c but I could be wrong Dave **= Email 16 ==========================** Date: Sun, 09 Jan 2005 11:28:18 -0800 From: "Dave Yeo" Subject: Re: emx\include\netdb.h:123: parse error before `int' On Sun, 9 Jan 2005 09:52:35 +0000, John Poltorak wrote: >I thought WGET was buildable on OS/2 straight out of the box... It was, really all these projects need someone monitoring CVS and the developers mailing list to catch these changes. A lot of the time its such a small fix and the developers are happy to change things for us. Dave **= Email 17 ==========================** Date: Sun, 9 Jan 2005 19:28:59 +0000 From: John Poltorak Subject: Re: Perl & CPAN.pm On Sun, Jan 09, 2005 at 11:16:30AM -0800, Dave Yeo wrote: > On Sun, 9 Jan 2005 12:00:41 +0000, John Poltorak wrote: > > >Does anyone else have any problems with using the CPAN module with Perl? > > I've used the CPAN module (though not for a year or more) and never had any problems with corrupt files. Did have some other problems eg make test failing which I had to cleanup by hand when installing some modules but otherwise it worked pretty well. This was with my own build of perl It's my own build of Perl, but the problem seems to be introduced from the way WGET is being run when retrieving files. For some reason '-O -' is included in the command line and I'm sure this is causing some newline problems, but I can't figure out where this additional parameter comes from... > Dave > -- John **= Email 18 ==========================** Date: Sun, 09 Jan 2005 11:25:59 -0800 From: "Dave Yeo" Subject: Re: Webmin and CPAN On Sun, 09 Jan 2005 13:20:12 +0000 (GMT), Lyn St George wrote: >In the meantime, you could see what you think of this version >running on my local OS/2 box, at >https://merlin.zolotek.net:10055, login ux2/ux2 (hopefully it's been >sufficiently defanged - I don't know who is lurking on this list ...) I tried this and get a connection refused message when first trying to connect :( Dave **= Email 19 ==========================** Date: Sun, 9 Jan 2005 19:45:39 +0000 From: John Poltorak Subject: Re: emx\include\netdb.h:123: parse error before `int' On Sun, Jan 09, 2005 at 11:28:18AM -0800, Dave Yeo wrote: > On Sun, 9 Jan 2005 09:52:35 +0000, John Poltorak wrote: > > >I thought WGET was buildable on OS/2 straight out of the box... > > It was, Can you remember which version? > really all these projects need someone monitoring CVS and the developers mailing list to catch these changes. A lot of the time its such a small fix and the developers are happy to change things for us. That would be an awful lot of lists to monitor, but I guess it might not be too onerous once we get up to date and establish a core list of apps to maintain. Also, I guess as apps mature they tend to stabilise and have fewer sets of revisions, or a longer development cycle... > Dave > -- John **= Email 20 ==========================** Date: Sun, 9 Jan 2005 19:48:04 +0000 From: John Poltorak Subject: Re: Re: Openssl-0.9.7e --5thEi1zFxNz+Kssz Content-Type: text/plain; charset=us-ascii On Sun, Jan 09, 2005 at 11:24:15AM -0800, Dave Yeo wrote: > On Sun, 9 Jan 2005 19:05:17 +0100, Stefan.Neis at t-online.de wrote: > > > > >Or maybe the long term plan of OpenSSL maintainers is to > >completely get rid of all those str(n)casecmp defines and always > >call OPENSSL_strncasecmp on all platforms, then o_str.c should > >be fixed to do the right thing on OS/2. > > I got the feeling that this is there long term plan and another ifelse or two should be added to o_str.c but I could be wrong I suggest adding the attached patch to OpenSSL v0.9.7e. Any have any objections to it? > Dave > -- John --5thEi1zFxNz+Kssz Content-Type: text/plain; charset=us-ascii Content-Disposition: attachment; filename="P:\\unixos2\\patches\\openssl-0.9.7e.diff" --- crypto/o_str.c~ Thu Jul 8 08:32:50 2004 +++ crypto/o_str.c Sun Jan 9 19:16:30 2005 at at -70,7 +70,7 at at int OPENSSL_strncasecmp(const char *str1, const char *str2, size_t n) { -#if defined(OPENSSL_SYS_VMS) +#if defined(OPENSSL_SYS_VMS) || defined(OPENSSL_SYS_OS2) while (*str1 && *str2 && n) { int res = toupper(*str1) - toupper(*str2); at at -94,7 +94,7 at at } int OPENSSL_strcasecmp(const char *str1, const char *str2) { -#if defined(OPENSSL_SYS_VMS) +#if defined(OPENSSL_SYS_VMS) || defined(OPENSSL_SYS_OS2) return OPENSSL_strncasecmp(str1, str2, (size_t)-1); #elif defined(OPENSSL_SYS_WINDOWS) return _stricmp(str1, str2); --5thEi1zFxNz+Kssz-- **= Email 21 ==========================** Date: Sun, 09 Jan 2005 15:05:31 -0500 From: lsunley at mb.sympatico.ca Subject: Tracking source changes - was - Re: emx\include\netdb.h:123: parse error before `int' What we need is a "build farm" setup. The postgreSQL guys have set up machines that automatically pull updates from CVS and do a standard ../configure, make, make install, make installcheck to verify that the code will build on the various platforms. It could be implemented as a set of .cmd files for each of the projects that have to be monitored, and everyone could volunteer for a subset of the apps that have to be tracked... Lorne In <20050109194539.K11877 at warpix.org>, on 01/09/05 at 07:45 PM, John Poltorak said: >On Sun, Jan 09, 2005 at 11:28:18AM -0800, Dave Yeo wrote: >> On Sun, 9 Jan 2005 09:52:35 +0000, John Poltorak wrote: >> >> >I thought WGET was buildable on OS/2 straight out of the box... >> >> It was, >Can you remember which version? >> really all these projects need someone monitoring CVS and the developers mailing list to catch these changes. A lot of the time its such a small fix and the developers are happy to change things for us. >That would be an awful lot of lists to monitor, but I guess it might not >be too onerous once we get up to date and establish a core list of apps >to maintain. Also, I guess as apps mature they tend to stabilise and >have fewer sets of revisions, or a longer development cycle... >> Dave >> -- ----------------------------------------------------------- lsunley at mb.sympatico.ca ----------------------------------------------------------- **= Email 22 ==========================** Date: Sun, 9 Jan 2005 22:22:05 +0100 From: Stefan.Neis at t-online.de Subject: Re: Re: Openssl-0.9.7e Hi, > I suggest adding the attached patch to OpenSSL v0.9.7e. > > Any have any objections to it? Yes. We really should use the version from the C library (which ideally is optimized "as far as possible" (or can be in the future) instead of using the homegrown replacement, that OpenSSL maintainers add for the platforms that don't have such a C library function. So, personally, I'd rather prefer the appended patch. Regards, Stefan ------------------------ *** o_str_orig.c Thu Jul 8 08:32:52 2004 --- o_str.c Sun Jan 9 22:13:10 2005 *************** *** 89,92 **** --- 89,94 ---- #elif defined(OPENSSL_SYS_WINDOWS) return _strnicmp(str1, str2, n); + #elif defined(OPENSSL_SYS_OS2) + return strnicmp(str1, str2, n); #else return strncasecmp(str1, str2, n); *************** *** 99,102 **** --- 101,106 ---- #elif defined(OPENSSL_SYS_WINDOWS) return _stricmp(str1, str2); + #elif defined(OPENSSL_SYS_OS2) + return stricmp(str1, str2); #else return strcasecmp(str1, str2); **= Email 23 ==========================** Date: Sun, 9 Jan 2005 22:29:53 +0100 From: Stefan.Neis at t-online.de Subject: Re: emx\include\netdb.h:123: parse error before `int' Hi, > > really all these projects need someone monitoring CVS and the developers mailing list to catch these changes. A lot of the time its such a small fix and the developers are happy to change things for us. > > That would be an awful lot of lists to monitor, Yes. I guess it's more practical to just try to compiled newly released versions and try to get potential bugs fixed in the next one. Monitoring all the relevant lists seems just impossible to me ... > but I guess it might not > be too onerous once we get up to date and establish a core list of apps to > maintain. Also, I guess as apps mature they tend to stabilise and have > fewer sets of revisions, or a longer development cycle... There always is a new esoteric feature that you can implement. If only 1% of the users even knows about it (and only 1% of that 1% actually thinks it might be useful sometime), who cares? I.e. in many cases that stabilisation just doesn't happen. Just consider word processing. Current apps are just as immature as 20 years ago and what new important feature does e.g. Word for Windows 2003 really offer when compared to Word (for DOS) 5.0? Regards, Stefan **= Email 24 ==========================** Date: Sun, 09 Jan 2005 14:15:04 -0800 From: Dave Yeo Subject: Re: Re: Openssl-0.9.7e On Sun, 2005-09-01 at 19:48 +0000, John Poltorak wrote: > On Sun, Jan 09, 2005 at 11:24:15AM -0800, Dave Yeo wrote: > > On Sun, 9 Jan 2005 19:05:17 +0100, Stefan.Neis at t-online.de wrote: > > > > > > > >Or maybe the long term plan of OpenSSL maintainers is to > > >completely get rid of all those str(n)casecmp defines and always > > >call OPENSSL_strncasecmp on all platforms, then o_str.c should > > >be fixed to do the right thing on OS/2. > > > > I got the feeling that this is there long term plan and another ifelse or two should be added to o_str.c but I could be wrong > > I suggest adding the attached patch to OpenSSL v0.9.7e. > > Any have any objections to it? > Isn't strncasecmp etc already undefined at this point? Dave **= Email 25 ==========================** Date: Sun, 09 Jan 2005 14:21:32 -0800 From: Dave Yeo Subject: Re: Tracking source changes - was - Re: On Sun, 2005-09-01 at 15:05 -0500, lsunley at mb.sympatico.ca wrote: > What we need is a "build farm" setup. The postgreSQL guys have set up > machines that automatically pull updates from CVS and do a standard > ./configure, make, make install, make installcheck to verify that the > code will build on the various platforms. > > It could be implemented as a set of .cmd files for each of the > projects > that have to be monitored, and everyone could volunteer for a subset > of > the apps that have to be tracked... > > Lorne > This is exactly what is needed with the volunteers also being subscribed to the developers mailing lists. I will volunteer some bandwidth and computer time, just have to think about which packages Dave **= Email 26 ==========================** Date: Sun, 09 Jan 2005 14:26:18 -0800 From: Dave Yeo Subject: Re: Re: Openssl-0.9.7e On Sun, 2005-09-01 at 22:22 +0100, Stefan.Neis at t-online.de wrote: > So, personally, I'd rather prefer the appended patch. > > Regards, > Stefan > > ------------------------ > > *** o_str_orig.c Thu Jul 8 08:32:52 2004 > --- o_str.c Sun Jan 9 22:13:10 2005 > *************** > *** 89,92 **** > --- 89,94 ---- > #elif defined(OPENSSL_SYS_WINDOWS) > return _strnicmp(str1, str2, n); > + #elif defined(OPENSSL_SYS_OS2) > + return strnicmp(str1, str2, n); > #else > return strncasecmp(str1, str2, n); > *************** > *** 99,102 **** > --- 101,106 ---- > #elif defined(OPENSSL_SYS_WINDOWS) > return _stricmp(str1, str2); > + #elif defined(OPENSSL_SYS_OS2) > + return stricmp(str1, str2); > #else > return strcasecmp(str1, str2); > > Have you tried this? I tried almost the identical patch last night and got some errors, can't remember exactly right now (rebooted into Linux) but IIRC a combination of parse errors and a complaint about a nonvoid function. Dave **= Email 27 ==========================** Date: Sun, 09 Jan 2005 16:35:50 -0500 From: lsunley at mb.sympatico.ca Subject: Re: Tracking source changes - was - Re: emx\include\netdb.h:123: parse error before `int' In <1105309292.3907.8.camel at amad>, on 01/09/05 at 02:21 PM, Dave Yeo said: >On Sun, 2005-09-01 at 15:05 -0500, lsunley at mb.sympatico.ca wrote: > What >we need is a "build farm" setup. The postgreSQL guys have set up > >machines that automatically pull updates from CVS and do a standard > >./configure, make, make install, make installcheck to verify that the > >code will build on the various platforms. >> >> It could be implemented as a set of .cmd files for each of the >> projects >> that have to be monitored, and everyone could volunteer for a subset >> of >> the apps that have to be tracked... >> >> Lorne >> >This is exactly what is needed with the volunteers also being subscribed >to the developers mailing lists. >I will volunteer some bandwidth and computer time, just have to think >about which packages >Dave I'll be setting up "build farms" this week for the postgreSQL code and the unixODBC code and the freetds ODBC driver. I'll probably do others that are used for the postgreSQL builds like openjade and opensp if I have the time. Lorne -- ----------------------------------------------------------- lsunley at mb.sympatico.ca ----------------------------------------------------------- **= Email 28 ==========================** Date: Mon, 10 Jan 2005 01:03:43 +0000 (GMT) From: "Lyn St George" Subject: Re: Webmin and CPAN On Sun, 9 Jan 2005 14:45:53 +0000, John Poltorak wrote: > >I'll revert back to an older version of Webmin if that works. I'm more >interested in having an older version of a working module than the latest >which needs fixing... I'm beginning to suspect the perl version might be the problem. I currently have 5.8.4, but I don't think it's ever been used in anger. Webmin prior to this would have been on perl 5.8.0 (IIRC) or even 5.6.1. I've now got v1.170, v1.140 and v1.121 running - I'm certain that 1.121 used to install cpan modules, but now it won't even display anything on the download.cgi page at all but just hangs there. 1.70 gets as far as 'failed to install module: download failed'. Great .. Which version of perl are you using? - Lyn **= Email 29 ==========================** Date: Mon, 10 Jan 2005 10:08:15 +0000 From: John Poltorak Subject: Re: Tracking source changes - was - Re: emx\include\netdb.h:123: parse error before `int' On Sun, Jan 09, 2005 at 04:35:50PM -0500, lsunley at mb.sympatico.ca wrote: > In <1105309292.3907.8.camel at amad>, on 01/09/05 > at 02:21 PM, Dave Yeo said: > > >> It could be implemented as a set of .cmd files for each of the > >> projects > >> that have to be monitored, and everyone could volunteer for a subset > >> of > >> the apps that have to be tracked... > >> > >> Lorne > >> > > >This is exactly what is needed with the volunteers also being subscribed > >to the developers mailing lists. > >I will volunteer some bandwidth and computer time, just have to think > >about which packages > >Dave > > I'll be setting up "build farms" this week for the postgreSQL code and the > unixODBC code and the freetds ODBC driver. I'll probably do others that > are used for the postgreSQL builds like openjade and opensp if I have the > time. I've heard the term 'build farm' in the past but never understood it, but sounds like a number of machines busily engaged in making compiles, in which case, I wonder if you have ever come across distcc...:- http://distcc.samba.org/ This sounds like the sort of thing which can speed up builds by spreading the load over several machines. A couple of us looked at trying to get it built on OS/2 several months ago but without any success, but it wouldn't hurt if a fresh pair of eyes had a browse through it. And if you have a spare farm, maybe you could take a look at UX2BS which is my attempt to create a standard build environment for Unix apps on OS/2. This is something which ought to be built from scratch every day or two to see ensure the update of any components co-exists harmoniously with the overall environment. enviro > Lorne > > > -- > ----------------------------------------------------------- > lsunley at mb.sympatico.ca > ----------------------------------------------------------- > -- John **= Email 30 ==========================** Date: Mon, 10 Jan 2005 10:13:52 +0000 From: John Poltorak Subject: Re: Re: Openssl-0.9.7e On Sun, Jan 09, 2005 at 02:26:18PM -0800, Dave Yeo wrote: > On Sun, 2005-09-01 at 22:22 +0100, Stefan.Neis at t-online.de wrote: > > So, personally, I'd rather prefer the appended patch. > > > > Regards, > > Stefan > > > > ------------------------ > > > > *** o_str_orig.c Thu Jul 8 08:32:52 2004 > > --- o_str.c Sun Jan 9 22:13:10 2005 > > *************** > > *** 89,92 **** > > --- 89,94 ---- > > #elif defined(OPENSSL_SYS_WINDOWS) > > return _strnicmp(str1, str2, n); > > + #elif defined(OPENSSL_SYS_OS2) > > + return strnicmp(str1, str2, n); > > #else > > return strncasecmp(str1, str2, n); > > *************** > > *** 99,102 **** > > --- 101,106 ---- > > #elif defined(OPENSSL_SYS_WINDOWS) > > return _stricmp(str1, str2); > > + #elif defined(OPENSSL_SYS_OS2) > > + return stricmp(str1, str2); > > #else > > return strcasecmp(str1, str2); > > > > > > Have you tried this? I tried almost the identical patch last night and > got some errors, can't remember exactly right now (rebooted into Linux) > but IIRC a combination of parse errors and a complaint about a nonvoid > function. It works for me. I've updated UX2BS, so if you run ux2_refresh then 'build openssl' it should work for you. If it doesn't, then let me know. > Dave > -- John **= Email 31 ==========================** Date: Mon, 10 Jan 2005 10:19:14 +0000 From: John Poltorak Subject: Re: Webmin and CPAN On Mon, Jan 10, 2005 at 01:03:43AM +0000, Lyn St George wrote: > On Sun, 9 Jan 2005 14:45:53 +0000, John Poltorak wrote: > > > > >I'll revert back to an older version of Webmin if that works. I'm more > >interested in having an older version of a working module than the latest > >which needs fixing... > > I'm beginning to suspect the perl version might be the problem. I currently > have 5.8.4, but I don't think it's ever been used in anger. Webmin prior > to this would have been on perl 5.8.0 (IIRC) or even 5.6.1. I've now got > v1.170, v1.140 and v1.121 running - I'm certain that 1.121 used to install > cpan modules, but now it won't even display anything on the download.cgi > page at all but just hangs there. 1.70 gets as far as 'failed to install module: > download failed'. Great .. > > Which version of perl are you using? I'm using v5.8.6 but have a problem with CPAN.pm which causes downloaded modules to get corrupted. Until I work out what is wrong there, I won't be able to use the cpan module in Webmin. > > - > Lyn > -- John **= Email 32 ==========================** Date: Mon, 10 Jan 2005 10:26:20 +0000 From: John Poltorak Subject: Re: distcc On Thu, Dec 30, 2004 at 12:43:26PM +0000, John Poltorak wrote: > On Thu, Dec 30, 2004 at 03:54:50AM -0800, Dave Yeo wrote: > > On Thu, 2004-30-12 at 09:52 +0000, John Poltorak wrote: > > > Has anyone ever used DISTCC (http://distcc.samba.org/) ? > > > > > > It seems like a pretty neat utility although I've never seen it in > > > practise. Is there any reason why it shouldn't work on OS/2? > > > > > > Didn't we go over this last year? > > The subject was brought up but I don't recall any conclusion as to whether > it would be possible to get it working on OS/2. Just thought I'd add that I just downloaded the most recent version:- http://distcc.samba.org/ftp/distcc/distcc-2.18.3.tar.bz2 and it actually built! but getting it setup looks as though it will be complicated... > > Dave > > -- John **= Email 33 ==========================** Date: Mon, 10 Jan 2005 12:05:11 +0100 From: Stefan.Neis at t-online.de Subject: Re: Re: Openssl-0.9.7e Hi, > > *** o_str_orig.c Thu Jul 8 08:32:52 2004 > > --- o_str.c Sun Jan 9 22:13:10 2005 > > *************** > > *** 89,92 **** > > --- 89,94 ---- > > #elif defined(OPENSSL_SYS_WINDOWS) > > return _strnicmp(str1, str2, n); > > + #elif defined(OPENSSL_SYS_OS2) > > + return strnicmp(str1, str2, n); > > #else > > return strncasecmp(str1, str2, n); > > *************** > > *** 99,102 **** > > --- 101,106 ---- > > #elif defined(OPENSSL_SYS_WINDOWS) > > return _stricmp(str1, str2); > > + #elif defined(OPENSSL_SYS_OS2) > > + return stricmp(str1, str2); > > #else > > return strcasecmp(str1, str2); > > > > > > Have you tried this? I tried almost the identical patch last night and > got some errors, can't remember exactly right now (rebooted into Linux) > but IIRC a combination of parse errors and a complaint about a nonvoid > function. Yes, I tried it, no problem over here (EMX/gcc-2.8.1). Actually, I only had completed the static build when sending that mail, but a couple of minutes later, the DLL build completed as well without problem. Regards, Stefan **= Email 34 ==========================** Date: Mon, 10 Jan 2005 11:21:08 +0000 From: John Poltorak Subject: Re: Re: Openssl-0.9.7e On Mon, Jan 10, 2005 at 12:05:11PM +0100, Stefan.Neis at t-online.de wrote: > > Have you tried this? I tried almost the identical patch last night and > > got some errors, can't remember exactly right now (rebooted into Linux) > > but IIRC a combination of parse errors and a complaint about a nonvoid > > function. > > Yes, I tried it, no problem over here (EMX/gcc-2.8.1). > Actually, I only had completed the static build when sending that mail, > but a couple of minutes later, the DLL build completed as well without > problem. It works fine for me too. The only thing left now is to make sure your patch is rolled into the official distribution so the v0.9.7f works out of the box (unless a new 'feature' has been added). Who should patches be sent to? I guess there isn't an official OS/2 maintainer... > Regards, > Stefan -- John **= Email 35 ==========================** Date: Mon, 10 Jan 2005 12:49:28 +0100 From: Stefan.Neis at t-online.de Subject: Re: Tracking source changes - was - Re: emx\include\netdb.h:123: parse error before `int' Hi, > I've heard the term 'build farm' in the past but never understood it, but > sounds like a number of machines busily engaged in making compiles, in > which case, I wonder if you have ever come across distcc...:- > > http://distcc.samba.org/ I think build farm rather means a number of machines each one building a different configuration or different apps or the same app for different operating systems, so there "normally" isn't much point in using distcc. > This sounds like the sort of thing which can speed up builds by spreading > the load over several machines. A couple of us looked at trying to get it > built on OS/2 several months ago but without any success, but it wouldn't > hurt if a fresh pair of eyes had a browse through it. Personally, I think it's much easier to just let different OS/2 machines build different packages rather than trying to distribute each package. Regards, Stefan **= Email 36 ==========================** Date: Mon, 10 Jan 2005 12:21:03 +0000 From: John Poltorak Subject: Re: Lowercase variable names On Sun, Jan 09, 2005 at 03:35:22PM +0100, Andreas Buening wrote: > John Poltorak wrote: > > > > How would I set variable names to lowercase in a cmd file? > > Not at all. > > > AIUI shell scripts can do this so maybe I can achieve this by running a > > shell script using EXTPROC. > > This can never work because any other shell would be executed as > a child process. Well this cmd file is what I came up for building Webmin:- extproc /bin/sh export perl=/usr/bin/perl.exe export allow=127.0.0.1 export port=10001 export login=admin export password=password export config_dir=/etc/webmin export var_dir=/var/log//webmin setup.sh /usr/local/webmin It seems to work OK. > > Or is there an alternative way of creating lowercase variable names? > > The only way is to start a Unix shell, to set the lower case variables > and finally to start another cmd within that shell. > > > Bye, > Andreas -- John **= Email 37 ==========================** Date: Mon, 10 Jan 2005 23:27:11 +1000 (EST) From: "Brian Havard" Subject: Re: Re: Openssl-0.9.7e On Sun, 9 Jan 2005 22:22:05 +0100, Stefan.Neis at t-online.de wrote: > Hi, > >> I suggest adding the attached patch to OpenSSL v0.9.7e. >> >> Any have any objections to it? > >Yes. We really should use the version from the C library >(which ideally is optimized "as far as possible" (or can >be in the future) instead of using the homegrown replacement, >that OpenSSL maintainers add for the platforms that don't >have such a C library function. > >So, personally, I'd rather prefer the appended patch. I agree, in fact I was preparing to submit an identical patch but I was held up by the fact that their rsync server is refusing connections. I wanted to diff against the latest 0.9.7-stable but can't update my CVS..... > ------------------------ > >*** o_str_orig.c Thu Jul 8 08:32:52 2004 >--- o_str.c Sun Jan 9 22:13:10 2005 >*************** >*** 89,92 **** >--- 89,94 ---- > #elif defined(OPENSSL_SYS_WINDOWS) > return _strnicmp(str1, str2, n); >+ #elif defined(OPENSSL_SYS_OS2) >+ return strnicmp(str1, str2, n); > #else > return strncasecmp(str1, str2, n); >*************** >*** 99,102 **** >--- 101,106 ---- > #elif defined(OPENSSL_SYS_WINDOWS) > return _stricmp(str1, str2); >+ #elif defined(OPENSSL_SYS_OS2) >+ return stricmp(str1, str2); > #else > return strcasecmp(str1, str2); -- ______________________________________________________________________________ | Brian Havard | "He is not the messiah! | | brianh at kheldar.apana.org.au | He's a very naughty boy!" - Life of Brian | ------------------------------------------------------------------------------