Date: Mon, 26 Apr 2004 00:04:20 EST-10EDT,10,-1,0,7200,3,-1,0,7200,3600 Subject: [UnixOS2_Archive] No. 360 ************************************************** Sunday 25 April 2004 Number 360 ************************************************** Subjects for today 1 Re: _lstat : Dave Yeo" 2 Re: _lstat : Dave Yeo" 3 Re: _lstat : John Poltorak 4 crypt.a : Dave Yeo" 5 Re: crypt.a : Yuri Dario" 6 Re: crypt.a : John Poltorak 7 ED tutorial : John Poltorak 8 Re: Autoconf 2.59 problem : nickk" 9 Re: Autoconf 2.59 problem : John Poltorak 10 Re: crypt.a : Dave Yeo" 11 Re: __getenv_findEnv : unresolved external : Stefan.Neis at t-online.de **= Email 1 ==========================** Date: Sat, 24 Apr 2004 08:04:19 -0800 From: "Dave Yeo" Subject: Re: _lstat On Sat, 24 Apr 2004 10:13:50 +0100, John Poltorak wrote: >On Fri, Apr 23, 2004 at 07:11:03PM -0800, Dave Yeo wrote: >> On Fri, 23 Apr 2004 22:39:26 +0100, John Poltorak wrote: >> >> >What can I do an undeined symbol _lstat? >> >> #ifdef __EMX__ >> #define lstat(n, b) stat(n, b) >> #endif /* EMX */ > >Actually, after digging through my archives I found the solution suggested >by Andreas the last time it happened - that was to run autoheader after >autoconf. This solution worked perfectly well with MKTEMP v1.4 when I last >tried it and also with the newer version (1.5) when I tried it yesterday. >The thing is it only works using the OS/2 version of Autoconf v2.50, not >the latest version. Maybe some experimental code was put in and then taken >out. From my point of view it's much more preferable to leave source code >alone if possible and am pleased that just running autoheader does the >trick, automagically, in this case. > > >> Dave > > >-- >John > > > **= Email 2 ==========================** Date: Sat, 24 Apr 2004 08:16:38 -0800 From: "Dave Yeo" Subject: Re: _lstat On Sat, 24 Apr 2004 10:13:50 +0100, John Poltorak wrote: >On Fri, Apr 23, 2004 at 07:11:03PM -0800, Dave Yeo wrote: >> On Fri, 23 Apr 2004 22:39:26 +0100, John Poltorak wrote: >> >> >What can I do an undeined symbol _lstat? >> >> #ifdef __EMX__ >> #define lstat(n, b) stat(n, b) >> #endif /* EMX */ > >Actually, after digging through my archives I found the solution suggested >by Andreas the last time it happened - that was to run autoheader after >autoconf. This solution worked perfectly well with MKTEMP v1.4 when I last >tried it and also with the newer version (1.5) when I tried it yesterday. >The thing is it only works using the OS/2 version of Autoconf v2.50, not >the latest version. Maybe some experimental code was put in and then taken >out. From my point of view it's much more preferable to leave source code >alone if possible and am pleased that just running autoheader does the >trick, automagically, in this case. > There must be code in there to handle systems without lstat. INteresting that you have to rerun autoheader for it to work. Where is the MKTEMP archive available? Dave **= Email 3 ==========================** Date: Sat, 24 Apr 2004 16:26:36 +0100 From: John Poltorak Subject: Re: _lstat On Sat, Apr 24, 2004 at 08:16:38AM -0800, Dave Yeo wrote: > On Sat, 24 Apr 2004 10:13:50 +0100, John Poltorak wrote: > > >Actually, after digging through my archives I found the solution suggested > >by Andreas the last time it happened - that was to run autoheader after > >autoconf. This solution worked perfectly well with MKTEMP v1.4 when I last > >tried it and also with the newer version (1.5) when I tried it yesterday. > >The thing is it only works using the OS/2 version of Autoconf v2.50, not > >the latest version. Maybe some experimental code was put in and then taken > >out. From my point of view it's much more preferable to leave source code > >alone if possible and am pleased that just running autoheader does the > >trick, automagically, in this case. > > > There must be code in there to handle systems without lstat. INteresting that you have to rerun autoheader for it to work. As I say, it works auto*magically*. Don't ask me how. Also it doesn't appear to work in recent Autoconfs. I used v2.50-r5 for OS/2. > Where is the MKTEMP archive available? It's obvious:- http://www.mktemp.org/ ;-) > Dave -- John **= Email 4 ==========================** Date: Sat, 24 Apr 2004 09:01:56 -0800 From: "Dave Yeo" Subject: crypt.a Anyone know where to get a version of libcrypt? Dave **= Email 5 ==========================** Date: Sat, 24 Apr 2004 18:39:57 +0200 (CDT) From: "Yuri Dario" Subject: Re: crypt.a Hi, >Anyone know where to get a version of libcrypt? check ufc on hobbes. Bye, Yuri Dario /* * member of TeamOS/2 - Italy * http://www.os2power.com/yuri * http://www.teamos2.it */ **= Email 6 ==========================** Date: Sat, 24 Apr 2004 18:58:20 +0100 From: John Poltorak Subject: Re: crypt.a On Sat, Apr 24, 2004 at 09:01:56AM -0800, Dave Yeo wrote: > Anyone know where to get a version of libcrypt? It's one of the first things that gets built when you install UX2BS. > Dave -- John **= Email 7 ==========================** Date: Sat, 24 Apr 2004 19:26:25 +0100 From: John Poltorak Subject: ED tutorial In case anyone wants to marvel at ED :-).... check out the ED tutorial here:- http://tinyurl.com/2aa6g I'm still trying to work out if the OS/2 port of GNU ed actually works... -- John **= Email 8 ==========================** Date: Sun, 25 Apr 2004 00:17:58 +0400 (MSD) From: "nickk" Subject: Re: Autoconf 2.59 problem On Sat, 24 Apr 2004 11:13:07 +0100, John Poltorak wrote: >There seems to be a problem with the OS/2 version of Autoconf which is not >present in the original distribution... > >This is the error I get when running configure on bc v1.06:- > > >checking for lib.h... no >checking for an ANSI C-conforming const... yes >checking for size_t... yes >checking for ptrdiff_t... yes >checking for vprintf... yes >checking for _doprnt... no >checking for isgraph... yes >checking for setvbuf... yes >./configure[5435]: syntax error: `if' unmatched I get such errors on almost every configure script generated by autoconf 2.59. **= Email 9 ==========================** Date: Sat, 24 Apr 2004 21:25:56 +0100 From: John Poltorak Subject: Re: Autoconf 2.59 problem On Sun, Apr 25, 2004 at 12:17:58AM +0400, nickk wrote: > On Sat, 24 Apr 2004 11:13:07 +0100, John Poltorak wrote: > > >There seems to be a problem with the OS/2 version of Autoconf which is not > >present in the original distribution... > > > >This is the error I get when running configure on bc v1.06:- > > > > > >checking for lib.h... no > >checking for an ANSI C-conforming const... yes > >checking for size_t... yes > >checking for ptrdiff_t... yes > >checking for vprintf... yes > >checking for _doprnt... no > >checking for isgraph... yes > >checking for setvbuf... yes > >./configure[5435]: syntax error: `if' unmatched > > I get such errors on almost every configure script generated by autoconf 2.59. Can you give me an example? It usually works for me, but there was an exception in one case. Have you tried installing the same version direct from GNU? It's available here:- ftp://ftp.gnu.org/pub/gnu/autoconf/autoconf-2.59.tar.gz -- John **= Email 10 ==========================** Date: Sat, 24 Apr 2004 18:24:22 -0800 From: "Dave Yeo" Subject: Re: crypt.a On Sat, 24 Apr 2004 18:58:20 +0100, John Poltorak wrote: >On Sat, Apr 24, 2004 at 09:01:56AM -0800, Dave Yeo wrote: >> Anyone know where to get a version of libcrypt? > >It's one of the first things that gets built when you install UX2BS. Not here, I don't have a (lib)crypt.a anywhere including my UX2BS tree. I now see thanks to Yuri's pointer that _crypt is in ufc.a which is installed with UX2BS. It's a X program, actually a X server, tightvnc which has a stripped down 3.3.2 Xserver. I'm using Holgers buildos2.cmd script and the imake process is asking for crypt.a. Anyways replacing crypt.a with ufc.a in vnc.def did the trick Dave **= Email 11 ==========================** Date: Sun, 25 Apr 2004 12:24:48 +0100 From: Stefan.Neis at t-online.de Subject: Re: __getenv_findEnv : unresolved external Hi, > Anyone know what I can do about:- ? > > __getenv_findEnv : unresolved external Probably something for the FAQ. That error message is the typical result of compiling against Posix/2 headers (or linking a library compiled with Posix/2) and not including Posix/2 libraries in the linking step. Regards, Stefan