From: UnixOS2 Archive
To: "UnixOS2 Archive" <UnixOS2_Archive at unixos2.com>
Date: Mon, 27 Jan 2003 04:49:11 EST-10EDT,10,-1,0,7200,3,-1,0,7200,3600
Subject: [UnixOS2_Archive] No. 26

**************************************************
Sunday 26 January 2003
 Number  26
**************************************************

Subjects for today
 
1   `long_options' : John Poltorak <jp at manninghammills.org>
2  Re:  Help Getting Perl to Run : John Poltorak <jp at manninghammills.org>
3  Re:  Help Getting Perl to Run : Hannes Hromadka <HromadkaJ at gmx.at>
4   termcap.lib : John Poltorak <jp at manninghammills.org>
5  Re:  `long_options' : John Poltorak <jp at manninghammills.org>
6  Re:  `long_options' : Stefan Neis <neis at cdc.informatik.tu-darmstadt.de>
7   Unix Under OS/2 : Hakan" <agents at meddatainc.com>
8  Re:  make question : Henry Sobotka <sobotka at axess.com>
9  Re:  Understanding EXEEXT : Maynard" <maynard.o at discourse.mentabolism.org>
10  Re:  Help Getting Perl to Run : John Poltorak <jp at manninghammills.org>
11  Re:  Help Getting Perl to Run : Hannes Hromadka <HromadkaJ at gmx.at>
12  Re:  Help Getting Perl to Run : Hannes Hromadka <HromadkaJ at gmx.at>
13  Re:  Perl - use: not found : Henry Sobotka <sobotka at axess.com>
14   make question : Dave Saville" <dave.saville at ntlworld.com>
15  Re:  UnixOS/2 docs : Jeff Robinson <jeffnik at anecho.mb.ca>
16   Understanding EXEEXT : John Poltorak <jp at manninghammills.org>
17   Perl - use: not found : John Poltorak <jp at manninghammills.org>
18  Re:  UnixOS/2 docs : John Poltorak <jp at manninghammills.org>
19  Re:  Understanding EXEEXT : John Poltorak <jp at manninghammills.org>
20  Re:  `long_options' : John Poltorak <jp at manninghammills.org>
21  Re:  wxWindows & varying build environments : Andreas Buening <andreas.buening at nexgo.de>
22   UnixOS/2 docs : Andreas Buening <andreas.buening at nexgo.de>
23  Re:  Understanding EXEEXT : Andreas Buening <andreas.buening at nexgo.de>
24  Re:  `long_options' : Thomas Hoffmann <thoffman at zappa.sax.de>
25  Re:  Understanding EXEEXT : John Poltorak <jp at manninghammills.org>
26  Re:  UnixOS/2 docs : Thomas Hoffmann <thoffman at zappa.sax.de>
27  Re:  `long_options' : John Poltorak <jp at manninghammills.org>
28  Re:  `long_options' : Andreas Buening <andreas.buening at nexgo.de>
29  Re:  Understanding EXEEXT : Andreas Buening <andreas.buening at nexgo.de>
30  Re:  UnixOS/2 docs : Andreas Buening <andreas.buening at nexgo.de>

**= Email   1 ==========================**

Date:  Mon, 27 Jan 2003 10:54:19 +0000
From:  John Poltorak <jp at manninghammills.org>
Subject:   `long_options'

Anyone familiar with `long_options' ?

I get this error when trying to build FILE v3.39:-


make  all-am
make[1]: Entering directory `U:/unixos2/workdir/file-3.39'
gcc -DHAVE_CONFIG_H -I. -I. -I.   -DMAGIC='"/usr/local/share/magic"'   -c 
`test -f 'file.c' || echo './'`file.c
In file included from file.h:48,
                 from file.c:28:
u:\emx\include\sys/stat.h:7: warning: #warning <sys/stat.h> requires 
<sys/types.h>
In file included from file.c:28:
file.h:101: parse error before `int32_t'
file.h:101: warning: no semicolon at end of struct or union
file.h:102: warning: data definition has no type or storage class
file.h:114: parse error before `}'
file.h:114: warning: empty declaration
file.c: In function `main':
file.c:131: elements of array `long_options' have incomplete type
file.c:133: warning: excess elements in struct initializer after `long_options[0]'
file.c:133: warning: excess elements in struct initializer after `long_options[0]'
...
...
file.c:149: warning: excess elements in struct initializer after `long_options[13]'
file.c:150: invalid use of undefined type `struct option'
file.c:131: storage size of `long_options' isn't known
make[1]: *** [file.o] Error 1


This is the relevant code from file.c:-


#define OPTSTRING	"bcdf:ikm:nsvzCL"
#ifdef HAVE_GETOPT_H
	int longindex;
	static struct option long_options[] =    /* <===== line 131  */
	{
		{"version", 0, 0, 'v'},
		{"help", 0, 0, 0},
		{"brief", 0, 0, 'b'},
		{"checking-printout", 0, 0, 'c'},
		{"debug", 0, 0, 'd'},
		{"files-from", 1, 0, 'f'},
		{"mime", 0, 0, 'i'},
		{"keep-going", 0, 0, 'k'},
#ifdef S_IFLNK
		{"dereference", 0, 0, 'L'},
#endif
		{"magic-file", 1, 0, 'm'},
		{"uncompress", 0, 0, 'z'},
		{"no-buffer", 0, 0, 'n'},
		{"special-files", 0, 0, 's'},
		{"compile", 0, 0, 'C'},
		{0, 0, 0, 0},
	};
#endif



OS/2 releases of FILE have always been made by Alexander Mai in the past, 
presumably after he has added some patch or other, but I have never seen 
any of these patches. Has anyone ever seen them?

I haven't had any reply from Alexander about this. Does anyone know if he 
has changed his address?


-- 
John



**= Email   2 ==========================**

Date:  Mon, 27 Jan 2003 11:47:35 +0000
From:  John Poltorak <jp at manninghammills.org>
Subject:  Re:  Help Getting Perl to Run

On Mon, Jan 27, 2003 at 12:06:35PM +0100, Hannes Hromadka wrote:
> On Fri, Jan 24, 2003 at 09:57:57PM -0500, Hakan wrote:
> 
> Hello:
> 
> May I add a "me too" to this mail ?
> 
> > [D:\perl\bin]testperl
> > ECHO is off.
> > Testing perl...  Getting version...
> > ECHO is off.
> > 
> > Process terminated by SIGILL
>  
> > Per the instructions with perl 5.7.2 "These binaries are highly
> > optimized (-O6) for Pentium Pro or equivalent (i686) machines, but in
> > theory should run on anything down to a 386."  I am trying to run it on
> > a Cyrix 586-100 which I understand has the Pentium instruction set but
> > behaves like a 486 and ought to run perl.  The machine has W4 + FP15
> > installed.
> 
> That could be the problem
> 
> Yesterday I tried perl 5.6.1 on a Pentium I with the same result 
> 
> SIGILL exception anytime I try to start perl.exe
> 
> It works with minor problems on a P II

If you, or anyone else is interested in running Perl v5.8.0, I'd suggest 
subscribing to the UX2BS mailing list where you will get a script for 
building it yourself. For details, see:-

http://powerusersbbs.net/mailman/listinfo/ux2bs

> 
>     Ciao,  Hannes
> 
> -- 
> Johannes Hromadka     | Email Office: HromadkaJ at gmx.at
>                       |       Home  : Johannes.Hromadka at gmx.net
> Vienna/Austria/Europe | OECC: http://www.oecc.org/
> >>> Rust never sleeps   (borrowed from Neil YOUNG) <<<

-- 
John



**= Email   3 ==========================**

Date:  Mon, 27 Jan 2003 12:06:35 +0100
From:  Hannes Hromadka <HromadkaJ at gmx.at>
Subject:  Re:  Help Getting Perl to Run

On Fri, Jan 24, 2003 at 09:57:57PM -0500, Hakan wrote:

Hello:

May I add a "me too" to this mail ?

> [D:\perl\bin]testperl
> ECHO is off.
> Testing perl...  Getting version...
> ECHO is off.
> 
> Process terminated by SIGILL
 
> Per the instructions with perl 5.7.2 "These binaries are highly
> optimized (-O6) for Pentium Pro or equivalent (i686) machines, but in
> theory should run on anything down to a 386."  I am trying to run it on
> a Cyrix 586-100 which I understand has the Pentium instruction set but
> behaves like a 486 and ought to run perl.  The machine has W4 + FP15
> installed.

That could be the problem

Yesterday I tried perl 5.6.1 on a Pentium I with the same result 

SIGILL exception anytime I try to start perl.exe

It works with minor problems on a P II



    Ciao,  Hannes

-- 
Johannes Hromadka     | Email Office: HromadkaJ at gmx.at
                      |       Home  : Johannes.Hromadka at gmx.net
Vienna/Austria/Europe | OECC: http://www.oecc.org/
>>> Rust never sleeps   (borrowed from Neil YOUNG) <<<

**= Email   4 ==========================**

Date:  Mon, 27 Jan 2003 12:21:57 +0000
From:  John Poltorak <jp at manninghammills.org>
Subject:   termcap.lib

EMX includes termcap.lib and btermcap.lib. I presume the first is built 
from GNU termcap and the other is based on BSD. How can I be sure?


Are there any programs I can try building which use the BSD library, and 
how would they specifically choose this version if both were present?


-- 
John


**= Email   5 ==========================**

Date:  Mon, 27 Jan 2003 13:07:39 +0000
From:  John Poltorak <jp at manninghammills.org>
Subject:  Re:  `long_options'

On Mon, Jan 27, 2003 at 01:26:52PM +0100, Stefan Neis wrote:
> On Mon, 27 Jan 2003, John Poltorak wrote:
> 
> > file.h:101: parse error before `int32_t'
> > file.h:101: warning: no semicolon at end of struct or union
> > file.h:102: warning: data definition has no type or storage class
> 
> I suppose, gcc is already totally confused at that point in the file,
> feel free to ignore the remaining error messages ...
> 
> > This is the relevant code from file.c:-
> 
> Probably not. :-(

You're right.

I just got a reply fromthe author who said:-

> -- Subject: File v3.39 for OS/2
>
> The problem is that you don't have inttypes.h ion OS2. This is a new thing,
> and Alexander (to my knowledge) was able to build file unmodified. You
> need to add the appropriate typedefs:
>
> i.e.
>
> typedef unsigned int uint32_t;
> typedef int int32_t;

Can you suggest what I need to do here?

Does Posix/2 have the required definitions?

I didn't see an inttypes.h when I looked. Where would I expect to find 
one?


> 	Regards,
> 		Stefan
> -- 
> Micro$oft is not an answer. It is a question. The answer is 'no'.
> 


-- 
John



**= Email   6 ==========================**

Date:  Mon, 27 Jan 2003 13:26:52 +0100 (CET)
From:  Stefan Neis <neis at cdc.informatik.tu-darmstadt.de>
Subject:  Re:  `long_options'

On Mon, 27 Jan 2003, John Poltorak wrote:

> file.h:101: parse error before `int32_t'
> file.h:101: warning: no semicolon at end of struct or union
> file.h:102: warning: data definition has no type or storage class

I suppose, gcc is already totally confused at that point in the file,
feel free to ignore the remaining error messages ...

> This is the relevant code from file.c:-

Probably not. :-(
> 
> #define OPTSTRI NG	"bcdf:ikm:nsvzCL"
> #ifdef HAVE_GETOPT_H
> 	int longindex;
> 	static struct option long_options[] =    /* <===== line 131  */

	Regards,
		Stefan
-- 
Micro$oft is not an answer. It is a question. The answer is 'no'.


**= Email   7 ==========================**

Date:  Mon, 27 Jan 2003 14:57:13 -0500 (EST)
From:  "Hakan" <agents at meddatainc.com>
Subject:   Unix Under OS/2

Fellow OS/2 enthusiasts,

If anyone has an interest in using the basic Unix application set under
OS/2, the Unix-OS/2 group has put together scripts which not only
download all files required from a server, but also builds the
applictions.  I used the bootstrap script myself and built perl last
night.

The bootstrap script is:

wget
ftp://unixos2: at 213.152.37.92/pub/unixos2/build_system/lib/ux2_bootstrap.
cmd

which requires that you already have wget (possibly also emx) and you
do need link386.exe (from the OS/2 CD) to build the applications. 
Everything else is downloaded.  Several new directories are created by
the script -- \bin, \etc, \home, \temp, \unixos2 and \usr -- all on the
disk of your choice.

The relevant Unix-OS/2 mailing lists are:

os2-unix at manninghammills.org for the environment
ux2bs-admin at powerusersbbs.net for the build system itself

Recommended!

Hakan



**= Email   8 ==========================**

Date:  Mon, 27 Jan 2003 14:57:22 -0500
From:  Henry Sobotka <sobotka at axess.com>
Subject:  Re:  make question

Dave Saville wrote:
> 
> As you can see that gcc line is *not* the one it should be using.
> These rules are seen and work OK on OS/2 & Solaris.
> 
> Ideas?

My guess is that make sees ALL as the target, but not the intermediate
object files, and so uses the implicit rule for making % from %.c in a
single compile-and-link step.

Try specifying the object files as intermediate targets (dependencies),
or invoking make with -r to disable the implicit rules altogether.

h~

**= Email   9 ==========================**

Date:  Mon, 27 Jan 2003 15:57:21 -0600 (CST)
From:  "Maynard" <maynard.o at discourse.mentabolism.org>
Subject:  Re:  Understanding EXEEXT

On Mon, 27 Jan 2003 21:08:19 +0000, John Poltorak wrote:

>So what do I do to automatically create patch.exe? 

can a scripts/pre-conf/ script patch the distributed configuration
file?

-- Maynard




**= Email   10 ==========================**

Date:  Mon, 27 Jan 2003 16:47:47 +0000
From:  John Poltorak <jp at manninghammills.org>
Subject:  Re:  Help Getting Perl to Run

On Mon, Jan 27, 2003 at 05:38:30PM +0100, Hannes Hromadka wrote:
> On Mon, Jan 27, 2003 at 11:47:35AM +0000, John Poltorak wrote:

> > If you, or anyone else is interested in running Perl v5.8.0, I'd suggest 
> > subscribing to the UX2BS mailing list where you will get a script for 
> > building it yourself. For details, see:-
> > 
> > http://powerusersbbs.net/mailman/listinfo/ux2bs
> 
> I subscribed, but if I access the archive I get an error 8-(
> 
> Archive File Not Found
> 
> No file /ux2bs/ (/unixos2/home/mailman/archives/private/ux2bs/) 

The mailing list is using Mailman which has only just been ported to OS/2 
and archiving is not yet working, but should be available before too long.

 
> So, please can you post the info here.


Since you have subscribed there, I'll try and provide the necessary info 
for you to get started. There are ongoing discussion about getting the 
best Perl build there, so there is no need to cross post to this list.
  
> 
>     Ciao,  Hannes
> 
> -- 
> Johannes Hromadka     | Email Office: HromadkaJ at gmx.at
>                       |       Home  : Johannes.Hromadka at gmx.net
> Vienna/Austria/Europe | OECC: http://www.oecc.org/
> >>> Rust never sleeps   (borrowed from Neil YOUNG) <<<


-- 
John



**= Email   11 ==========================**

Date:  Mon, 27 Jan 2003 17:21:19 +0100
From:  Hannes Hromadka <HromadkaJ at gmx.at>
Subject:  Re:  Help Getting Perl to Run

On Mon, Jan 27, 2003 at 11:47:35AM +0000, John Poltorak wrote:
 
> If you, or anyone else is interested in running Perl v5.8.0, I'd suggest 
> subscribing to the UX2BS mailing list where you will get a script for 
> building it yourself. For details, see:-

I tried, but make failed somewhere while compiling the extensions.

> http://powerusersbbs.net/mailman/listinfo/ux2bs

one mailinglist more 

    Ciao,  Hannes

-- 
Johannes Hromadka     | Email Office: HromadkaJ at gmx.at
                      |       Home  : Johannes.Hromadka at gmx.net
Vienna/Austria/Europe | OECC: http://www.oecc.org/
>>> Rust never sleeps   (borrowed from Neil YOUNG) <<<

**= Email   12 ==========================**

Date:  Mon, 27 Jan 2003 17:38:30 +0100
From:  Hannes Hromadka <HromadkaJ at gmx.at>
Subject:  Re:  Help Getting Perl to Run

On Mon, Jan 27, 2003 at 11:47:35AM +0000, John Poltorak wrote:
>  
> > > Per the instructions with perl 5.7.2 "These binaries are highly
> > > optimized (-O6) for Pentium Pro or equivalent (i686) machines, but in
> > 
> > That could be the problem
> > 
> > Yesterday I tried perl 5.6.1 on a Pentium I with the same result 
> > 
> > SIGILL exception anytime I try to start perl.exe

Is this package really compiled for Pentuim II or newer processors ?

> If you, or anyone else is interested in running Perl v5.8.0, I'd suggest 
> subscribing to the UX2BS mailing list where you will get a script for 
> building it yourself. For details, see:-
> 
> http://powerusersbbs.net/mailman/listinfo/ux2bs

I subscribed, but if I access the archive I get an error 8-(

Archive File Not Found

No file /ux2bs/ (/unixos2/home/mailman/archives/private/ux2bs/) 

So, please can you post the info here.
 

    Ciao,  Hannes

-- 
Johannes Hromadka     | Email Office: HromadkaJ at gmx.at
                      |       Home  : Johannes.Hromadka at gmx.net
Vienna/Austria/Europe | OECC: http://www.oecc.org/
>>> Rust never sleeps   (borrowed from Neil YOUNG) <<<

**= Email   13 ==========================**

Date:  Mon, 27 Jan 2003 17:50:40 -0500
From:  Henry Sobotka <sobotka at axess.com>
Subject:  Re:  Perl - use: not found

John Poltorak wrote:
> 
> ./help2man --include=./help2man.h2m \
>     --output=help2man.1 ./help2man
> ./help2man[24]: use: not found
> ./help2man[25]: use: not found
> ./help2man[26]: use: not found
> ./help2man[27]: syntax error: `(' unexpected
> make: *** [help2man.1] Error 1

This is the result of sh trying to run a Perl script. You either have to
fix the makefile to run "perl help2man", or (have help2man.PL) insert

eval 'exec perl -S $0 ${1+"$ at "}'
   if $running_under_some_shell;

right after the shebang for OS/2.

h~

**= Email   14 ==========================**

Date:  Mon, 27 Jan 2003 18:34:38 +0000 (GMT)
From:  "Dave Saville" <dave.saville at ntlworld.com>
Subject:   make question

Hi gang

I am trying to get some of my command line stuff to compile on OS/2,
Solaris & Red Hat linux.

I can get it to work on OS/2 with GNU Make version 3.76.1

I can get it to work on Solaris with whatever make it has - make -v
does not work so I guess its not gnu make

But on linux it not only won't work it's not using my rules. Make is
GNU Make version 3.79.1

Here is the make file. the only difference between them are the paths
and that the os/2 ones end in .exe

Here is the linux makefile:

.SILENT:

CC = gcc 

HOME = /home/db/src

ALL : blanks \
calander \
hexdmpr \
modify \
pager \
quicksort \
rpcping \
split \
timesync \
wipe
	 at echo utilities are up to date

% : %.o
	$(CC) $*.o -o $* \
	-L$(HOME)/subs -lsocket -lnsl -lsubs
#	cp -u $* d:\bin
	 at echo $* compiled
%.o : %.c
	$(CC) -D__LINUX__ -c $< \
	-I$(HOME)/subs

make -f makefile_linux
gcc      calander.c   -o calander
calander.c:11:18: subs.h: No such file or directory

As you can see that gcc line is *not* the one it should be using.
These rules are seen and work OK on OS/2 & Solaris. 

Ideas?


-- 
Regards

Dave Saville


**= Email   15 ==========================**

Date:  Mon, 27 Jan 2003 19:35:30 -0600
From:  Jeff Robinson <jeffnik at anecho.mb.ca>
Subject:  Re:  UnixOS/2 docs

Andreas Buening wrote:
> John Poltorak wrote:
> 
>>On Mon, Jan 27, 2003 at 09:46:26PM +0100, Andreas Buening wrote:
> 
> 
> [UnixOS/2 docs]
> 
> 
>>>package       latest stable   latest beta     Unix version    NLS     restrictions
>>>sed   4.0.5           4.0.5           4.0.5           yes     none
>>>
>>>"sed" could be a link to the "sed.html" page which lists extra infos.
>>>Whether special flags are needed to compile it, whether the test suite
>>>works, a short description (e.g. from the manpage).
>>
>>Andreas,
>>
>>Could we incorporate this with the UnixOS/2 Build System?
> 
> 
> I don't see how this can help for the Build System. I think
> we need some central source (preferably on unixos2.com) where
> the current state is documented. Which program is in which
> state? Which shortcomings? Restrictions? Special Build flags?
> Maintainer? What has to be done? I'd really like to get the
> info within 5 seconds if I don't remember the state of e.g. perl.
> I know Jeff is maintaining the UnixOS/2 web pages but he might
> need some input by the developers. :-)
> 
> [snip]
> 

Indeed, I'd very much appreciate help!  I'd like input from folks 
that're working on different ports as I generally can't remember what is 
going on half the time between this list and dabbling with my own projects.

I uploaded a new set of pages last night (I don't think they've been 
published yet) that fill out a little bit more information, as well as 
making links to the UnixOS2 Build System.  My ultimate goal would be to 
find a way to "automate" the pages a bit more, or make it so people 
working in certain areas may maintain their own information.  Hmmm... 
should really bug Ian to see if he'll try Zope...

If anyone is working on projects they want listed, please drop me a line 
and I'll try to include the info.

Thanks,

Jeff


-- 
----------------
Whatza JamochaMUD?
http://jamochamud.anecho.mb.ca

Or other stuff: http://www.anecho.mb.ca/~jeffnik
  -----------------------------------------------------------

  

**= Email   16 ==========================**

Date:  Mon, 27 Jan 2003 20:19:21 +0000
From:  John Poltorak <jp at manninghammills.org>
Subject:   Understanding EXEEXT


I'm trying to understand how the EXEEXT mechanism works under Autoconf...

I often see Makefile.in containing the line:-


EXEEXT =  at EXEEXT at 

which generally finds its way into Makefile, getting translated along the 
way as:-

EXEEXT = .exe

However in the case of some apps such as PATCH, this conversion does not 
get made and the resultant target for Make is patch rather than patch.exe,
which means an additional step is required.

What is it which resolves this variable, and why does it fail to work 
sometimes?

--
John




**= Email   17 ==========================**

Date:  Mon, 27 Jan 2003 20:49:03 +0000
From:  John Poltorak <jp at manninghammills.org>
Subject:   Perl - use: not found


When I try building HELP2MAN, I get the following error:-


creating ./config.status
creating Makefile
make
rm -f help2man
u:/usr/bin/perl.exe help2man.PL --stdout >help2man
chmod 555 help2man
./help2man --include=./help2man.h2m \
    --output=help2man.1 ./help2man
./help2man[24]: use: not found
./help2man[25]: use: not found
./help2man[26]: use: not found
./help2man[27]: syntax error: `(' unexpected
make: *** [help2man.1] Error 1


These are the lines in question and look perfectly ordinary to me:-


use 5.005;
use strict;
use Getopt::Long;
use Text::Tabs qw(expand);
use POSIX qw(strftime setlocale LC_TIME);


This looks like some sort of Perl setup problem... Any idea what's
missing?


-- 
John



**= Email   18 ==========================**

Date:  Mon, 27 Jan 2003 21:03:22 +0000
From:  John Poltorak <jp at manninghammills.org>
Subject:  Re:  UnixOS/2 docs

On Mon, Jan 27, 2003 at 09:46:26PM +0100, Andreas Buening wrote:
> Hello!
> 
> What do you think about adding some additional information to the
> "Current status" page (http://unix.os2site.com/pages/Status.html)?
> I'm thinking about separating the (not really) huge list of apps
> into different subsets of UnixOS/2 (Core/Standard/Docs/Developer/Misc)
> which contain lists (one line per package) of the currrent state
> (last stable version / last beta / latest Unix version / support level)
> 
> The first line of the "Core.html" subset could e.g. be
> 
> package	latest stable	latest beta	Unix version	NLS	restrictions
> sed	4.0.5		4.0.5		4.0.5		yes	none
> 
> "sed" could be a link to the "sed.html" page which lists extra infos.
> Whether special flags are needed to compile it, whether the test suite
> works, a short description (e.g. from the manpage).


Andreas,

Could we incorporate this with the UnixOS/2 Build System?

We are currently having reasonable success building numerous apps using a 
standardised build environment. There is a mailing list for people testing 
this environment, and your presence would definitely help speed things up.

You can get further details about the list here:-

http://powerusersbbs.net/mailman/listinfo/ux2bs

We already have around 30 apps which build reasonably well straight out of 
the box, although a few have minor problems. The aim is to get the built 
and installed completely without any user intervention at all. In some 
instances there are minor failures, but they may be due to some problem 
with autoconf or automake, and you seem to be the only person who knows 
the details of how these programs work. Today I tried the latest GROFF 
without any patcches or changes and it almost worked completely.

We are getting very close to a an excellent build framework and just need 
a few minor adjustments. Your help would be very much appreciated.
 
> 
> Btw. John, is os2-unix at manninghammills.org now supposed to
> be the home of the UnixOS/2 mailing list?


It is for the time being since I was forced to change my domain name...
although I may be changing it again in a few weeks time.
 
> 
> Bye,
> Andreas
> 
> -- 
> One OS to rule them all, One OS to find them,
> One OS to bring them all and in the darkness bind them
> In the Land of Mordor where the Shadows lie.


-- 
John


**= Email   19 ==========================**

Date:  Mon, 27 Jan 2003 21:08:19 +0000
From:  John Poltorak <jp at manninghammills.org>
Subject:  Re:  Understanding EXEEXT

On Mon, Jan 27, 2003 at 09:50:55PM +0100, Andreas Buening wrote:

> > What is it which resolves this variable, and why does it fail to work
> > sometimes?
> 
> It is generated by current versions of automake and is supposed
> to be replaced by ".exe" by the configure script. But this
> happens only for configure scripts created by autoconf 2.5xy
> and newer. And it doesn't help if the maintainer of the package
> doesn't make any use of $(EXEEXT) which seems to be the case
> for patch.

So what do I do to automatically create patch.exe? 

Should it work correctly if I rebuild configure using autoconf v2.57?

 
> 
> Bye,
> Andreas
> 
> -- 
> One OS to rule them all, One OS to find them,
> One OS to bring them all and in the darkness bind them
> In the Land of Mordor where the Shadows lie.


-- 
John



**= Email   20 ==========================**

Date:  Mon, 27 Jan 2003 21:17:27 +0000
From:  John Poltorak <jp at manninghammills.org>
Subject:  Re:  `long_options'

On Mon, Jan 27, 2003 at 09:53:14PM +0100, Thomas Hoffmann wrote:
> John Poltorak wrote:
> >>The problem is that you don't have inttypes.h ion OS2. This is a new thing,
> >>and Alexander (to my knowledge) was able to build file unmodified. You
> >>need to add the appropriate typedefs:
> ....
> >>typedef unsigned int uint32_t;
> >>typedef int int32_t;
> ....
> > Can you suggest what I need to do here?
> use Posix/2 (see below)
> 
> > Does Posix/2 have the required definitions?
> Yes, but they are hidden a bit:
> G:\thoffman\work\posix2.patched\libext\include>grep -r uint32 *
> .....
> i386/types.h:typedef    unsigned int             uint32_t;
> 
> --> You will get it via an include chain from unistd.h

So if I was using Posix/2 headers, then presumable configure would locate 
it automatically and 

 
> > I didn't see an inttypes.h when I looked. Where would I expect to find 
> > one?
> C9x says:
> 7.8 Format conversion of integer types <inttypes.h>
> 1 The header <inttypes.h> includes the header <stdint.h> and extends it
> with
> additional facilities provided by hosted implementations.
> 
> So, one could look for an inttypes.h on the 'net and adapt it to OS/2 (I 
> have no clue what the relation to the Posix std is, although)

Actually, I was told wrongly... It should have been stdint.h rather than 
inttypes.h. Both exist in GLIBC but I couldn't get things to work 
correctly - there were were a great many additional headers required by 
the included headers and I was going round in circles.

Maybe I'll see how far I get using the Posix/2 headers...

> Thomas.

-- 
John
 


**= Email   21 ==========================**

Date:  Mon, 27 Jan 2003 21:23:38 +0100
From:  Andreas Buening <andreas.buening at nexgo.de>
Subject:  Re:  wxWindows & varying build environments

Stefan Neis wrote:

[snip]

> - older versions of GNUmake supported "MAKE_SHELL", in the current
>   version, that variable is named "MAKESHELL" (no more underscore!).

Never heard of that name. Which make version did ever use
MAKE_SHELL?

[snip]

Bye,
Andreas

-- 
One OS to rule them all, One OS to find them,
One OS to bring them all and in the darkness bind them
In the Land of Mordor where the Shadows lie.

**= Email   22 ==========================**

Date:  Mon, 27 Jan 2003 21:46:26 +0100
From:  Andreas Buening <andreas.buening at nexgo.de>
Subject:   UnixOS/2 docs

Hello!

What do you think about adding some additional information to the
"Current status" page (http://unix.os2site.com/pages/Status.html)?
I'm thinking about separating the (not really) huge list of apps
into different subsets of UnixOS/2 (Core/Standard/Docs/Developer/Misc)
which contain lists (one line per package) of the currrent state
(last stable version / last beta / latest Unix version / support level)

The first line of the "Core.html" subset could e.g. be

package	latest stable	latest beta	Unix version	NLS	restrictions
sed	4.0.5		4.0.5		4.0.5		yes	none

"sed" could be a link to the "sed.html" page which lists extra infos.
Whether special flags are needed to compile it, whether the test suite
works, a short description (e.g. from the manpage).


Btw. John, is os2-unix at manninghammills.org now supposed to
be the home of the UnixOS/2 mailing list?


Bye,
Andreas

-- 
One OS to rule them all, One OS to find them,
One OS to bring them all and in the darkness bind them
In the Land of Mordor where the Shadows lie.

**= Email   23 ==========================**

Date:  Mon, 27 Jan 2003 21:50:55 +0100
From:  Andreas Buening <andreas.buening at nexgo.de>
Subject:  Re:  Understanding EXEEXT

John Poltorak wrote:
> 
> I'm trying to understand how the EXEEXT mechanism works under Autoconf...
> 
> I often see Makefile.in containing the line:-
> 
> EXEEXT =  at EXEEXT at 

[snip]

> What is it which resolves this variable, and why does it fail to work
> sometimes?

It is generated by current versions of automake and is supposed
to be replaced by ".exe" by the configure script. But this
happens only for configure scripts created by autoconf 2.5xy
and newer. And it doesn't help if the maintainer of the package
doesn't make any use of $(EXEEXT) which seems to be the case
for patch.


Bye,
Andreas

-- 
One OS to rule them all, One OS to find them,
One OS to bring them all and in the darkness bind them
In the Land of Mordor where the Shadows lie.

**= Email   24 ==========================**

Date:  Mon, 27 Jan 2003 21:53:14 +0100
From:  Thomas Hoffmann <thoffman at zappa.sax.de>
Subject:  Re:  `long_options'

John Poltorak wrote:
>>The problem is that you don't have inttypes.h ion OS2. This is a new thing,
>>and Alexander (to my knowledge) was able to build file unmodified. You
>>need to add the appropriate typedefs:
....
>>typedef unsigned int uint32_t;
>>typedef int int32_t;
....
> Can you suggest what I need to do here?
use Posix/2 (see below)

> Does Posix/2 have the required definitions?
Yes, but they are hidden a bit:
G:\thoffman\work\posix2.patched\libext\include>grep -r uint32 *
.....
i386/types.h:typedef    unsigned int             uint32_t;

--> You will get it via an include chain from unistd.h

> I didn't see an inttypes.h when I looked. Where would I expect to find 
> one?
C9x says:
7.8 Format conversion of integer types <inttypes.h>
1 The header <inttypes.h> includes the header <stdint.h> and extends it
with
additional facilities provided by hosted implementations.

So, one could look for an inttypes.h on the 'net and adapt it to OS/2 (I 
have no clue what the relation to the Posix std is, although)

Thomas.


**= Email   25 ==========================**

Date:  Mon, 27 Jan 2003 22:12:28 +0000
From:  John Poltorak <jp at manninghammills.org>
Subject:  Re:  Understanding EXEEXT

On Mon, Jan 27, 2003 at 03:57:21PM -0600, Maynard wrote:
> On Mon, 27 Jan 2003 21:08:19 +0000, John Poltorak wrote:
> 
> >So what do I do to automatically create patch.exe? 
> 
> can a scripts/pre-conf/ script patch the distributed configuration
> file?


Patches are meant to be applied from the \unixos2\patches directory 
normally, using this nifty code:-


PATCHFILE=$BLD_HOME/patches/$ARCHIVE.diff

if test -f $PATCHFILE; then
{

  patch -p`awk 'BEGIN {
      min_p = 1234567
    }

    $1 == "+++" || $1 == "***" {
      gsub (ARCHIVE ".*", "", $2)
      gsub (/[^/]/, "", $2)
      if (length($2) < min_p) min_p = length($2)
    }

  END {
    print min_p + 1
  }' ARCHIVE=$ARCHIVE $PATCHFILE` < $PATCHFILE

}
fi

So if there is a patch for an app it will be applied automatically...

Unfortunately in the case of PATCH a patch (this gets confusing...) has to 
be applied after configure runs but before Make. Ideally, autoconf should 
be able create the configure script which in turn will generate the 
Makefile as we want it. I don't know whether there is a bug in autoconf 
which is preventing the Makefile getting created properly...

 
> -- Maynard


-- 
John


**= Email   26 ==========================**

Date:  Mon, 27 Jan 2003 22:34:10 +0100
From:  Thomas Hoffmann <thoffman at zappa.sax.de>
Subject:  Re:  UnixOS/2 docs

I would like this very much. Looks like a user friendly "as much 
information as possible" solution. One could even point to other ports 
and mention the differences, list remaining nonconformancies etc.

Thomas.

Andreas Buening wrote:
> Hello!
> 
> What do you think about adding some additional information to the
> "Current status" page (http://unix.os2site.com/pages/Status.html)?
> I'm thinking about separating the (not really) huge list of apps
> into different subsets of UnixOS/2 (Core/Standard/Docs/Developer/Misc)
> which contain lists (one line per package) of the currrent state
> (last stable version / last beta / latest Unix version / support level)
> 
> The first line of the "Core.html" subset could e.g. be
> 
> package	latest stable	latest beta	Unix version	NLS	restrictions
> sed	4.0.5		4.0.5		4.0.5		yes	none
> 
> "sed" could be a link to the "sed.html" page which lists extra infos.
> Whether special flags are needed to compile it, whether the test suite
> works, a short description (e.g. from the manpage).
> 
> 
> Btw. John, is os2-unix at manninghammills.org now supposed to
> be the home of the UnixOS/2 mailing list?
> 
> 
> Bye,
> Andreas
> 


**= Email   27 ==========================**

Date:  Mon, 27 Jan 2003 22:48:56 +0000
From:  John Poltorak <jp at manninghammills.org>
Subject:  Re:  `long_options'

On Mon, Jan 27, 2003 at 11:41:35PM +0100, Andreas Buening wrote:
> John Poltorak wrote:
> 
> [inttypes.h]
> 
> > Actually, I was told wrongly... It should have been stdint.h rather than
> > inttypes.h. Both exist in GLIBC but I couldn't get things to work
> > correctly - there were were a great many additional headers required by
> > the included headers and I was going round in circles.
> 
> You can't use glibc headers directly. glibc has a different
> internal philosophy.
> 
> 
> > Maybe I'll see how far I get using the Posix/2 headers...
> 
> May I restate that we need _one_ _single_ set of consistent
> header files?

Yes, I understand that, but my prime concern was with getting the app 
built successully without any code changes. Once I know it does work then 
I can look for an optimum solution.

Besides, there may be missing functions in Posix/2 so it may have lead to 
the need to add something...
 
> 
> Bye,
> Andreas
> 
> -- 
> One OS to rule them all, One OS to find them,
> One OS to bring them all and in the darkness bind them
> In the Land of Mordor where the Shadows lie.

-- 
John



**= Email   28 ==========================**

Date:  Mon, 27 Jan 2003 23:41:35 +0100
From:  Andreas Buening <andreas.buening at nexgo.de>
Subject:  Re:  `long_options'

John Poltorak wrote:

[inttypes.h]

> Actually, I was told wrongly... It should have been stdint.h rather than
> inttypes.h. Both exist in GLIBC but I couldn't get things to work
> correctly - there were were a great many additional headers required by
> the included headers and I was going round in circles.

You can't use glibc headers directly. glibc has a different
internal philosophy.


> Maybe I'll see how far I get using the Posix/2 headers...

May I restate that we need _one_ _single_ set of consistent
header files?


Bye,
Andreas

-- 
One OS to rule them all, One OS to find them,
One OS to bring them all and in the darkness bind them
In the Land of Mordor where the Shadows lie.

**= Email   29 ==========================**

Date:  Mon, 27 Jan 2003 23:41:44 +0100
From:  Andreas Buening <andreas.buening at nexgo.de>
Subject:  Re:  Understanding EXEEXT

John Poltorak wrote:
> 
> On Mon, Jan 27, 2003 at 09:50:55PM +0100, Andreas Buening wrote:
> 
> > > What is it which resolves this variable, and why does it fail to work
> > > sometimes?
> >
> > It is generated by current versions of automake and is supposed
> > to be replaced by ".exe" by the configure script. But this
> > happens only for configure scripts created by autoconf 2.5xy
> > and newer. And it doesn't help if the maintainer of the package
> > doesn't make any use of $(EXEEXT) which seems to be the case
> > for patch.
> 
> So what do I do to automatically create patch.exe?
> 
> Should it work correctly if I rebuild configure using autoconf v2.57?

No, you have to patch (one of) the Makefile.am file(s).
For patch.exe maybe one line might be sufficient but there
might be other shortcomings, too.


Bye,
Andreas

-- 
One OS to rule them all, One OS to find them,
One OS to bring them all and in the darkness bind them
In the Land of Mordor where the Shadows lie.

**= Email   30 ==========================**

Date:  Mon, 27 Jan 2003 23:42:00 +0100
From:  Andreas Buening <andreas.buening at nexgo.de>
Subject:  Re:  UnixOS/2 docs

John Poltorak wrote:
> 
> On Mon, Jan 27, 2003 at 09:46:26PM +0100, Andreas Buening wrote:

[UnixOS/2 docs]

> > package       latest stable   latest beta     Unix version    NLS     restrictions
> > sed   4.0.5           4.0.5           4.0.5           yes     none
> >
> > "sed" could be a link to the "sed.html" page which lists extra infos.
> > Whether special flags are needed to compile it, whether the test suite
> > works, a short description (e.g. from the manpage).
> 
> Andreas,
> 
> Could we incorporate this with the UnixOS/2 Build System?

I don't see how this can help for the Build System. I think
we need some central source (preferably on unixos2.com) where
the current state is documented. Which program is in which
state? Which shortcomings? Restrictions? Special Build flags?
Maintainer? What has to be done? I'd really like to get the
info within 5 seconds if I don't remember the state of e.g. perl.
I know Jeff is maintaining the UnixOS/2 web pages but he might
need some input by the developers. :-)

[snip]


Bye,
Andreas

-- 
One OS to rule them all, One OS to find them,
One OS to bring them all and in the darkness bind them
In the Land of Mordor where the Shadows lie.