From: UnixOS2 Archive
To: "UnixOS2 Archive" <UnixOS2_Archive at unixos2.com>
Date: Mon, 22 Jul 2002 04:33:09 EST-10EDT,10,-1,0,7200,3,-1,0,7200,3600
Subject: [UnixOS2_Archive] No. 280

**************************************************
Sunday 21 July 2002
 Number  280
**************************************************

Subjects for today
 
1  gettext specification (was:  Pre-requisites for building GETTEXT) : Andreas Buening <andreas.buening at nexgo.de>
2   Next Apps : Maynard" <maynard.o at discourse.mentabolism.org>
3  Re:  the UnixOS2 ftp site : Maynard" <maynard.o at discourse.mentabolism.org>
4  Re:  ux2_bootstrap (revised) : John Poltorak <jp at eyup.org>
5  Re:  Clean Environment - Success : John Poltorak <jp at eyup.org>
6  Re:  the UnixOS2 ftp site : John Poltorak <jp at eyup.org>
7  Re:  Perl test results summary : Lyn St George" <lyn at zolotek.net>
8  Re:  UnixOS/2 bootstrap : John Poltorak <jp at eyup.org>
9  Re:  Next Apps : John Poltorak <jp at eyup.org>
10  Re:  ux2_bootstrap (revised) : Maynard" <maynard.o at discourse.mentabolism.org>
11  Re:  ux2_bootstrap (revised) : John Poltorak <jp at eyup.org>
12  Re:  ux2_bootstrap (revised) : Maynard" <maynard.o at discourse.mentabolism.org>
13  Re:  ux2_bootstrap (revised) : Maynard" <maynard.o at discourse.mentabolism.org>
14  Re:  ux2_bootstrap (revised) : xyzyx" <xyzyx at inbox.lv>
15  Re:  ux2_bootstrap (revised) : Maynard" <maynard.o at discourse.mentabolism.org>
16   ux2_bootstrap/Perl build results : xyzyx" <xyzyx at inbox.lv>
17  Re:  ux2_bootstrap (revised) : John Poltorak <jp at eyup.org>
18   Perl - Rexx line break failure : Sebastian Wittmeier (ShadoW)" <Sebastian.Wittmeier at ginko.de>
19  Re:  ux2_bootstrap/Perl build results : Maynard" <maynard.o at discourse.mentabolism.org>
20  Re:  Perl - Rexx line break failure : Sebastian Wittmeier (ShadoW)" <Sebastian.Wittmeier at ginko.de>
21   which SED : Maynard" <maynard.o at discourse.mentabolism.org>
22  Re:  ux2_bootstrap/Perl build results : xyzyx" <xyzyx at inbox.lv>

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

Date:  Mon, 22 Jul 2002 01:14:06 +0200
From:  Andreas Buening <andreas.buening at nexgo.de>
Subject:  gettext specification (was:  Pre-requisites for building GETTEXT)

John Poltorak wrote:
> 
> On Sun, Jul 21, 2002 at 05:13:22PM +0200, Andreas Buening wrote:
> > John Poltorak wrote:
> > >
> > > Can anyone tell me what are the pre-requisites to building GETTEXT?
> >
> > Oh, that's an easy (tm) question. ;-)
> > Of course, you need the basic survival kit (sh, perl, sed, make
> > and others).

[snip]

> I just want to know I need to be able to build an INTL.DLL based on
> Gettext 0.11.2 which will work with every utility which needs it.
> 
> I just want a script which works, where I don't have to press enter twice
> or re-run make unless it is the third Thursday of the month and there is a
> full moon :-)...

Okay, in this case. With a few modifications it is possible
to get configure running (without java and other stuff).
Currently I'm checking the code. There is one question left:

Normally gettext gets the current codepage from the OS/2 API,
but XFree86/2 uses latin1 regardless of the OS/2 codepage
(at least on my computer it's latin1 which is the Linux
default). This leads to the following effect:

If you run "gettext --help" in an OS/2 window you get
the help messages in your native language in your OS/2
codepage (e.g. cp850).
If you run "gettext --help" in an xterm window you get the
help messages also in your native language but gettext
prints the characters as cp850 while XFree86 displays latin1:
your accented characters are broken.

Possible solutions:
1) Find out at runtime whether the process runs in an xterm
   (how?) and which codepage XFree86/2 is using (again: how?)

2) Don't care about nls in xterms. Assume that the user has
   set OUTPUT_CHARSET correctly in XFree. This doesn't work
   if an X11 application has been started from a OS/2 window.

3) If the current process has no direct access to the keyboard
   assume it is running in an xterm and do:
   a) either no character conversion (assume the nls messages
      are already in the correct charset). This doesn't work
      if the process has been started from an OS/2 window
      without using "detach".
      .
   b) read correct codepage from /usr/lib/charset.alias
      (current value is cp850, but this could be changed)
      Problem: This is a fixed value and won't work for all
      languages.


As I have no idea how to implement (1), I prefer (3a).
If we don't care about nls in xterm then we don't need to
care about gettext at all (my opinion). Therefore (2)
is no solution. (3b) is not flexible enough. Any opinions?


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   2 ==========================**

Date:  Mon, 22 Jul 2002 06:06:09 -0500 (CDT)
From:  "Maynard" <maynard.o at discourse.mentabolism.org>
Subject:   Next Apps

On Mon, 22 Jul 2002 10:16:50 +0100, John Poltorak wrote:

>The apps I'd like to try building initially are BYACC, FLEX, GZIP, BZIP2, 
>LESS, AUTOCONF, AUTOMAKE, and SED and M4.

I'd like to recommend adding GREP and TAR, both of which don't behave
properly in their current binary form on os/2

In time,

--
Maynard



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

Date:  Mon, 22 Jul 2002 06:08:01 -0500 (CDT)
From:  "Maynard" <maynard.o at discourse.mentabolism.org>
Subject:  Re:  the UnixOS2 ftp site

On Mon, 22 Jul 2002 10:16:50 +0100, John Poltorak wrote:

>This script should be all you need:-
>
>wget ftp://unixos2: at eyup.org/pub/unixos2/build_system/lib/ux2_bootstrap.cmd

Thank you. I'll give it another whirl and report as, or better than,
usual.

If this file were in /pub/unixos2 with a README it would be a good
thing IMHO. I'll work on the README ;-}

--
Maynard



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

Date:  Mon, 22 Jul 2002 08:48:56 +0100
From:  John Poltorak <jp at eyup.org>
Subject:  Re:  ux2_bootstrap (revised)

On Sun, Jul 21, 2002 at 04:04:39PM -0500, xyzyx wrote:

> Looking at the batch files, it appears ux2_bootstrap.cmd sets the drive
> to %bldrt% and none of the other scripts change that. So the
> emx_inst.cmd is unzipping to the current drive? Which would be why the
> files are in the wrong place.

Yes, you are right. I obviously did not test things where uxrt and bldrt 
were on different drives. Sorry about that.

> Absolutely, X: and Y: were completely empty drives before starting. I
> made d:\unixos2 which contains nothing but the ux2_bootstrap.cmd and
> set it up the rest of the way with TVFS:
> 
> md d:\unixos2\bldrt
> md d:\unixos2\uxrt
> tvmount x:
> tvmount y:
> tvlink -rw -d x:\ D:\unixos2\bldrt
> tvlink -rw -d y:\ D:\unixos2\uxrt
> 
> I then ran ux2_bootstrap.cmd with the variables changed to:
> 
> set bldrt=x:
> set uxrt=y:
> set osrt=c:
> 
> This all look OK?

This is fine. What you need is to change emx_inst.cmd to include the line:-

%uxrt%

after setlocal.


I'll make that change and test it today to make sure it works.

 
> Thanks again,
> Paul


-- 
John



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

Date:  Mon, 22 Jul 2002 09:13:18 +0100
From:  John Poltorak <jp at eyup.org>
Subject:  Re:  Clean Environment - Success

On Sun, Jul 21, 2002 at 03:07:30PM -0500, Maynard wrote:
> 
> I'm declaring a successful completion of *_my_* search for the minimum environment required for building perl 5.8.0 and minimizing errors reported in harness.
> I did not bother trying to remove ETC, OS2_SHELL, PROMPT, TZ;
> TEMP is required; TMP is not;
> I'm not confident that I have consistent results regarding the two HOST vars. More testing might be able to drop one or both of these.
> 
> The environment inherited from OS/2 is:

I'd be interested to know what additional failures occur without any 
variables being inherited...

I thinks it important to know which variables are used and by which 
applications.
 
> ETC=N:\MPTN\ETC
> HOSTNAME=arcturus
> OS2_SHELL=N:\OS2\CMD.EXE

Any idea what uses this?


> PROMPT=[$p]

????

> TEMP=y:\temp

I didn't think Unix uses TEMP... AIUI, it's a DOSism. Unix uses TMP and 
also TMPDIR sometimes. 


> TZ=CST6CDT,4,1,0,7200,10,-1,0,7200,3600
> USE_HOSTS_FIRST=1





> --
> Maynard


-- 
John



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

Date:  Mon, 22 Jul 2002 10:16:50 +0100
From:  John Poltorak <jp at eyup.org>
Subject:  Re:  the UnixOS2 ftp site

On Sun, Jul 21, 2002 at 08:29:47PM -0500, Maynard wrote:
> John,
> 
> I suggest that your ftp site directory /pub/unixos2/ contain the
> necessary startup script[s] such as 'bootstrap'; and maybe the only
> other required tool (awget).
> 
> I suggest that we need some clarification on the distinction between
> "baseline" and "build_system".

Originally, these were two completely seperate sets of scripts, one for 
setting up a baseline system and the other was the actual build system for 
compiling apps, but I have now integrated everything under the build 
system. It may be an idea to keep baseline as an example of what is 
actually created after successfully running the full build... 

 
> It's probably time to organize also into the structure you announced
> several days ago:

> /unixos2/archives/baseline/emx
> /unixos2/archives/source

These are directories created locally in the process of building the 
baseline. There is no need to have them on the ftp server. Doing so is 
likely to generate extra traffic unnecessarily because people are likely 
to do recursive WGETs.



> /unixos2/bootstrap.cmd [per suggestion above]
> 
> 
> /unixos2/lib



> /unixos2/makefiles
> /unixos2/patches
> /unixos2/scripts
> /unixos2/variables


These may get slightly rearranged...


> /unixos2/logs

This should only be produced locally, although I do think if is useful to 
have a log to compare against, when something doesn't work right, but I 
wouldn't want people to grab it by default as it isn't required for 
building anything.


> 
> or whatever.
> 
> If the /pub/unixos2/ base has only one or two .cmd files, and maybe a
> README, then future visits would be more understandable than my current
> experience ;-}

This script should be all you need:-

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

In theory ux2_bootstrap will only retreive any updates after the first 
time it gets run, but there probably need to be some checks to prevent 
Perl getting built everytime, if a perfectly good Perl build is already in 
place. 
 
> If you'll announce when the ftp site is ready for the next testing, I'm
> ready to clear out my %uxrt% path and reconstruct local
> baseline/build_system.

Well, I think the basic build is more or less in place now, and Perl seems 
to build pretty well and consistantly for a number of people, so I  guess  
it's time to throw a few apps at it to see how well the build system copes  
with building them from source.

The apps I'd like to try building initially are BYACC, FLEX, GZIP, BZIP2, 
LESS, AUTOCONF, AUTOMAKE, and SED and M4.

Once I can get GETTEXT built, that will open the door to quite a few more.


 
> Thanks again,
> 
> --
> Maynard


-- 
John


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

Date:  Mon, 22 Jul 2002 10:32:00 +0000
From:  "Lyn St George" <lyn at zolotek.net>
Subject:  Re:  Perl test results summary

On Sat, 20 Jul 2002 18:12:51 +0100, John Poltorak wrote:

>> SW: lib/rx_cmprt.t             255 65280    18    3  16.67%  16-18
>
>How Ly avoids this I don't know.
>
>What's your secret Lyn? :-)...

Just twigged - this box is running Object Rexx; it's probably a 
fair guess that the failures are running on Classic Rexx.
-
Cheers
Lyn St George
+---------------------------------------------------------------------------------
+ http://www.zolotek.net .. eCommerce hosting, consulting
+ http://www.os2docs.org .. some 'How To' stuff ...
+----------------------------------------------------------------------------------


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

Date:  Mon, 22 Jul 2002 11:16:17 +0100
From:  John Poltorak <jp at eyup.org>
Subject:  Re:  UnixOS/2 bootstrap

On Sat, Jul 20, 2002 at 06:25:22PM -0500, xyzyx wrote:

> Last time I tried TVFS it caused all sorts of problems... crashing
> alot, etc.  Are there any [free] TVFS-like alternatives?

If you have a network you may be able to set up an alias such a UNIXROOT, 
so all you need to do is:-

net use u: unixroot

Another option is to use NFS, or even SRVIFS.

If you have large ammounts of RAM, you can set up one or more RAM disks 
using RAMFS.  


> thanks,
> paul


-- 
John



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

Date:  Mon, 22 Jul 2002 12:22:56 +0100
From:  John Poltorak <jp at eyup.org>
Subject:  Re:  Next Apps

On Mon, Jul 22, 2002 at 06:06:09AM -0500, Maynard wrote:
> On Mon, 22 Jul 2002 10:16:50 +0100, John Poltorak wrote:
> 
> >The apps I'd like to try building initially are BYACC, FLEX, GZIP, BZIP2, 
> >LESS, AUTOCONF, AUTOMAKE, and SED and M4.
> 
> I'd like to recommend adding GREP and TAR, both of which don't behave
> properly in their current binary form on os/2

GREP uses INTL therefore GETTEXT is a prerequisite.

I would like to build INTL from source rather than just downloading it 
from somewhere.

TAR, unfortunately, has proved unbuildable so far.

A number of people would be interested in getting it built and possibly 
updated but the source is incomplete and the OS/2 maintainer has 
disappeared.

The only usable source is for an old  16-bit app. I don't know if getting 
that built would serve any useful purpose. Maybe it would, I'm not sure...

 
> In time,
> 
> --
> Maynard


-- 
John



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

Date:  Mon, 22 Jul 2002 14:38:41 -0500 (CDT)
From:  "Maynard" <maynard.o at discourse.mentabolism.org>
Subject:  Re:  ux2_bootstrap (revised)

On Mon, 22 Jul 2002 08:48:56 +0100, John Poltorak wrote:

>This is fine. What you need is to change emx_inst.cmd to include the line:-
>
>%uxrt%
>
>after setlocal.

Maybe not. Notice in ux2_inst.cmd after calling emx_inst.cmd the line:
move \usr\dll\* \usr\lib

which I believe is in %uxrt%

Furthermore, files are extracted into %uxrt%\usr\dll by
baseline_inst.cmd which could just as easily put all those files into
\user\lib in the first place and skip the subsequent move.

Wherever they go, BEGINLIBPATH need to follow them.

I'm also finding errors in the newly distributed SED reporting as
version 2.05
Error occurs in the second half of emx...long.cmd
--------paste:------------
Patching cpp\algo.h                                  
SYS0002: The system cannot find the file specified.  
                                                     
 "GNURX"                                             
sed failed!                                          
-------end paste----------

replacing SED with v3.02 from the earlier archives results in success
here.

I'll double check this after I work out the drive letters.

--
Maynard



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

Date:  Mon, 22 Jul 2002 16:07:27 +0100
From:  John Poltorak <jp at eyup.org>
Subject:  Re:  ux2_bootstrap (revised)

On Tue, Jul 23, 2002 at 12:54:13AM +0100, Michael Taylor wrote:
> The file config.site had the path c:/bin/sh.exe hardcoded in it.

Yes, that does need to be changed. Thanks for pointing it out.


> Maybe because of the failure of sed? c:/bin on my system has 16bit
> dos executables.
> 
> It looks like the replacement sed I used is unhappy in the configure script
> e.g.
> Finding dependencies for mg.obj.
> sed: too many {'s
> Finding dependencies for reentr.obj.
> sed: too many {'s
> Finding dependencies for perl.obj.
> 
> Yes - it is unhappy:
> [E:\unixos2\workdir\perl-5.8.0]make
> `sh  cflags "optimize='-O2 -fomit-frame-pointer -malign-loops=2 -malign-jumps=2
> -malign-functions=2 -s'" malloc.obj` -Zdll malloc.c
>            CCCMD =  gcc -DPERL_CORE -c -Zomf -Zmt -DDOSISH -DOS2=2 -DEMBED -I. -D
> _EMX_CRT_REV_=64 -O2 -fomit-frame-pointer -malign-loops=2 -malign-jumps=2 -malig
> n-functions=2 -s -Wall
> In file included from malloc.c:244:
> perl.h:771: conflicting types for `sys_errlist'
> e:\usr\include\stdlib.h:147: previous declaration of `sys_errlist'
> make: *** [malloc.obj] Error 1
> 
> What is going on with the version of sed download by ux2_bootstrap?

I've used a backlevel version of SED in the baseline build, although it is 
one of the first apps I want to update through the build system.

However the version in the baseline build has been OK as far a building 
Perl goes. You may have had a problem with an incompatible GNURX.DLL or 
something like that.


 
> -- 
> Regards,
> Mick
> 
> --
> -----------------------------------------------------------------------
>      Michael Taylor                                 miket at pcug.org.au
>     -----------------------------------------------------------------
>               Home Page:  http://users.bigpond.net.au/miket5au
>     -----------------------------------------------------------------
> 
> -----------------------------------------------------------------------


-- 
John



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

Date:  Mon, 22 Jul 2002 16:16:13 -0500 (CDT)
From:  "Maynard" <maynard.o at discourse.mentabolism.org>
Subject:  Re:  ux2_bootstrap (revised)

Are we regressing here??
in emx_inst.cmd:

Are you sure that this is what you want to do??
mv emx/lib usrù
mv emx/include usrù
ù
You moved these lines from ux2_inst.cmd to emx_inst.cmd:
move \usr\dll\* \usr\libù
rd \usr\dllù

I just ran a test which extracted everything directly into \usr\lib and
see no reason to complain or sign of complaint. [changed in
baseline_inst.cmd and emx_inst.cmd /usr/dll to /usr/lib]

Perl Config fails (can't find compiler; aborts; script continues with
the rest of the process concluding with harness test failing 8 scripts
and 69/68656 subtests; probably not all that bad for being unconfigged,
but not what we want anyway. 

--sigh




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

Date:  Mon, 22 Jul 2002 16:47:06 -0500 (CDT)
From:  "Maynard" <maynard.o at discourse.mentabolism.org>
Subject:  Re:  ux2_bootstrap (revised)

On Mon, 22 Jul 2002 16:16:13 -0500 (CDT), Maynard wrote:

>Are you sure that this is what you want to do??
>mv emx/lib usrù
>mv emx/include usrù
>ù
>You moved these lines from ux2_inst.cmd to emx_inst.cmd:
>move \usr\dll\* \usr\libù
>rd \usr\dllù

Oh maybe I get it now ;-}

You use \usr\dll temporarily because "mv emx/lib usr" fails if /usr/lib
already exists;

silly me





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

Date:  Mon, 22 Jul 2002 19:24:40 -0500 (CDT)
From:  "xyzyx" <xyzyx at inbox.lv>
Subject:  Re:  ux2_bootstrap (revised)

On Mon, 22 Jul 2002 21:11:20 +0100, John Poltorak wrote:

>Here are a couple of revised files:-
>
>wget ftp://unixos2: at 213.152.37.92/pub/unixos2/build_system/lib/emx_inst.cmd
>wget ftp://unixos2: at 213.152.37.92/pub/unixos2/build_system/lib/ux2_inst.cmd

It now seems to work using these files (had to pause ux2_bootstrap and
copy these over since it pulled down the older ones). The whole
download/unzip/patch/emxomf process went fine, but it died when trying
to configure Perl:


Configure[36]: can't create temporary file /tmp/sh00981.000: No such
file or directory
Configure[41]: can't create temporary file /tmp/sh00981.001: No such
file or directory
Configure[142]: can't create temporary file /tmp/sh00981.002: No such
file or directory
Configure[1323]: can't create temporary file /tmp/sh00981.003: No such
file or directory
SYS1041: The name .\sharp is not recognized as an
internal or external command, operable program or batch file.
Hmm... '#!y:/usr/bin/sh' does not guarantee sh startup...
You may have to fix up the shell scripts to make sure y:/usr/bin/sh
runs them.
Configure[1343]: can't create temporary file /tmp/sh00981.004: No such
file or directory
Configure[1359]: can't create temporary file /tmp/sh00981.005: No such
file or directory
Configure[1359]: can't create temporary file /tmp/sh01026.006: No such
file or directory
Configure[1359]: can't create temporary file /tmp/sh00981.006: No such
file or directory
Configure[1359]: can't create temporary file /tmp/sh01033.007: No such
file or directory
Configure[1359]: can't create temporary file /tmp/sh00981.007: No such
file or directory
Configure[1359]: can't create temporary file /tmp/sh01040.008: No such
file or directory
Configure[1408]: can't create temporary file /tmp/sh00981.008: No such
file or directory
Configure[1430]: can't create temporary file /tmp/sh00981.009: No such
file or directory
Say 'sh Configure', not 'sh <Configure'


Everything after configure obviously fails. There is a /tmp dir is on
UXRT (y:), but not on BLDRT (x:). TMP=y:\tmp is apparently not
important to configure which looks like it's going to /tmp regardless
of the environment variable.

I created x:\tmp and tried the ''sed "s/^/build /" build.lst | cmd /q''
again and it seemed to be OK...

So I guess either %BLDRT% should have a /tmp on it, or fix the
configure to look for %TMP% instead of assuming /tmp

Anyway, now Perl is building and I'll report back the results later
(this is a 266MHz so it might be a while...)

Thanks,
Paul

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

Date:  Mon, 22 Jul 2002 20:09:01 -0500 (CDT)
From:  "Maynard" <maynard.o at discourse.mentabolism.org>
Subject:  Re:  ux2_bootstrap (revised)

Paul,

>Anyway, now Perl is building and I'll report back the results later
>(this is a 266MHz so it might be a while...

I log each step of the build process with 2>&1 | tee \logs\....
and am seeing the current build_system hide some errors during the
Configuration of perl.

[x:\logs]grep -c WHOA config.w
105      <-- current; not good

[x:\logs]grep -c WHOA config.2
5     <-- yesterday was good

Regards,
-- Maynard




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

Date:  Mon, 22 Jul 2002 21:01:24 -0500 (CDT)
From:  "xyzyx" <xyzyx at inbox.lv>
Subject:   ux2_bootstrap/Perl build results

OK, after the aforementioned changes, ux2 installed and tries to
compile Perl, here are the results:

Failed 6/726 test scripts, 99.17% okay. 36/68687 subtests failed,
99.95% okay.
Failed Test               Stat Wstat Total Fail  Failed  List of
Failed
------------------------------------------------------------------------
-------
../lib/ExtUtils/t/basic.t    1   256    17    1   5.88%  14
lib/os2_ea.t                            21    8  38.10%  7-11 14-16
lib/os2_process.t           10  2560   227   10   4.41%  80 85 94 171
174 181
                                                         186 196 206
209
lib/os2_process_kid.t                  227   13   5.73%  80 85 94 146
151 161
                                                         171 174 181
186 196
                                                         206 209
lib/rx_cmprt.t             255 65280    18    3  16.67%  16-18
op/stat.t                               73    1   1.37%  44
60 tests and 563 subtests skipped.


Regards,
Paul

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

Date:  Mon, 22 Jul 2002 21:11:20 +0100
From:  John Poltorak <jp at eyup.org>
Subject:  Re:  ux2_bootstrap (revised)

On Mon, Jul 22, 2002 at 02:38:41PM -0500, Maynard wrote:
> On Mon, 22 Jul 2002 08:48:56 +0100, John Poltorak wrote:
> 
> >This is fine. What you need is to change emx_inst.cmd to include the line:-
> >
> >%uxrt%
> >
> >after setlocal.
> 
> Maybe not. Notice in ux2_inst.cmd after calling emx_inst.cmd the line:
> move \usr\dll\* \usr\lib
> 
> which I believe is in %uxrt%
> 
> Furthermore, files are extracted into %uxrt%\usr\dll by
> baseline_inst.cmd which could just as easily put all those files into
> \user\lib in the first place and skip the subsequent move.

As you may expect, there's always more to it once you look into it.

Here are a couple of revised files:-

wget ftp://unixos2: at 213.152.37.92/pub/unixos2/build_system/lib/emx_inst.cmd
wget ftp://unixos2: at 213.152.37.92/pub/unixos2/build_system/lib/ux2_inst.cmd


 
> Wherever they go, BEGINLIBPATH need to follow them.
> 
> I'm also finding errors in the newly distributed SED reporting as
> version 2.05
> Error occurs in the second half of emx...long.cmd
> --------paste:------------
> Patching cpp\algo.h                                  
> SYS0002: The system cannot find the file specified.  
>                                                      
>  "GNURX"                                             
> sed failed!                                          
> -------end paste----------
> 
> replacing SED with v3.02 from the earlier archives results in success
> here.


It has been working with SED 2.05 here. You should find GNURX.DLL under 
%uxrt%\usr\dll if it has been extracted correctly, but it gets moved into 
\usr\lib eventually.

 
> Maynard

-- 
John


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

Date:  Mon, 22 Jul 2002 22:02:36 +0200 (CEST)
From:  "Sebastian Wittmeier (ShadoW)" <Sebastian.Wittmeier at ginko.de>
Subject:   Perl - Rexx line break failure

Hi,
I tried to solve rx_cmprt.t failing. I hope that exec_in_REXX_with() in
\os2\OS2\REXX\REXX.xs is the right place to change (the trivial
solution would be to modify the test file, but there is Rexx for other
systems, e.g. Regina).

The following diff was successful. It allocates memory with a borrowed
id (-> should be corrected). I tried to modify the string directly, but
I got SIGSEGVs.

Every \n is changed to \r.

diff -Naur old/REXX.xs new/REXX.xs
--- old/REXX.xs	Sat Jun  1 17:03:34 2002
+++ new/REXX.xs	Mon Jul 22 20:17:52 2002
 at  at  -73,6 +73,8  at  at 
     SV *res;
     char *subs = 0;
     int n = c;
+    int i;
+    char *cmd2;
 
     incompartment++;
 
 at  at  -84,8 +86,15  at  at 
 	    subs[n] = 1;
     }
 
+    Newz(728, cmd2, strlen(cmd)+1, char);
+    for(i=0;i<strlen(cmd); i++)
+      if(cmd[i]==10)
+        cmd2[i]=13;
+      else
+        cmd2[i]=cmd[i];
+
     MAKERXSTRING(args[0], NULL, 0);
-    MAKERXSTRING(inst[0], cmd,  strlen(cmd));
+    MAKERXSTRING(inst[0], cmd2,  strlen(cmd2));
     MAKERXSTRING(inst[1], NULL, 0);
     MAKERXSTRING(result,  NULL, 0);
     rc = pRexxStart(0, args,		/* No arguments */
 at  at  -103,6 +112,7  at  at 
     while (n--)
 	if (!subs[n])
 	    pRexxDeregisterFunction(handlerNames[n]);
+    Safefree(cmd2);
     if (c)
 	Safefree(subs);
 #if 0					/* Do we want to restore
these? */

Sebastian



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

Date:  Mon, 22 Jul 2002 22:28:40 -0500 (CDT)
From:  "Maynard" <maynard.o at discourse.mentabolism.org>
Subject:  Re:  ux2_bootstrap/Perl build results

On Mon, 22 Jul 2002 21:01:24 -0500 (CDT), xyzyx wrote:

>Failed 6/726 test scripts, 99.17% okay. 36/68687 subtests failed,

previously reported results in this neighborhood:

Failed 4/726 test scripts, 99.45% okay. 10/68689
Failed 3/726 test scripts, 99.59% okay.  5/68681
Failed 4/726 test scripts, 99.45% okay.  8/68656
Failed 6/726 test scripts, 99.17% okay. 10/68642
Failed 3/726 test scripts, 99.59% okay.  4/68471


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

Date:  Mon, 22 Jul 2002 22:36:08 +0200 (CEST)
From:  "Sebastian Wittmeier (ShadoW)" <Sebastian.Wittmeier at ginko.de>
Subject:  Re:  Perl - Rexx line break failure

Hi,

I should explain better:

Perl can execute Rexx commands. It passes them to the Rexx DLLs
installed by the operating system. In one test there is a line break in
the middle of the string, which holds the Rexx commands. And it was a
Unix line break (\n) instead of the OS/2 (\r\n). The OS/2 Rexx DLLs (at
least the ones from classic Rexx in opposite to the ones from Object
Rexx) only understand (\r) and (\r\n).
\n means ascii sign number 10 and is LF (line feed)
\r means ascii sign number 13 and is CR (carriage return)

I looked for a place, where the Rexx commands are given to the
operating system.

I found out (through testing) that I may not alter the parameter
holding the Rexx command (memory protection).
So I had to allocate my own memory and copy+alter the old parameter.
To allocate memory you need (with the revised memory allocation scheme,
which we already get to know earlier with the malloc problem) an id,
which identifies the request for memory. It is used to solve detected
memory leaks.
I don't know how to get such an unique id. So I borrowed another one
(doesn't affect the function of the program).

That patch should be redone by the OS/2 maintainer. It's only a
(working) demonstration how it can be done.

Sebastian



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

Date:  Mon, 22 Jul 2002 22:41:57 -0500 (CDT)
From:  "Maynard" <maynard.o at discourse.mentabolism.org>
Subject:   which SED

John,

What is it about v2.05 which has you prefer it over v3.02?
v2 requires the additional file which is a 'negative';
what does it have in the postitive ?

I have yet to achieve any successful baseline installs which use v2;
installs using v3 are fine. You must have some reason for liking v2
which I have not come across yet.

--Maynard


On Mon, 22 Jul 2002 21:11:20 +0100, John Poltorak wrote:

>> I'm also finding errors in the newly distributed SED reporting as
>> version 2.05
>> Error occurs in the second half of emx...long.cmd
>> --------paste:------------
>> Patching cpp\algo.h                                  
>> SYS0002: The system cannot find the file specified.  
>>                                                      
>>  "GNURX"                                             
>> sed failed!                                          
>> -------end paste----------
>> 
>> replacing SED with v3.02 from the earlier archives results in success
>> here.
>
>
>It has been working with SED 2.05 here. You should find GNURX.DLL under 
>%uxrt%\usr\dll if it has been extracted correctly, but it gets moved into 
>\usr\lib eventually.



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

Date:  Mon, 22 Jul 2002 23:58:47 -0500 (CDT)
From:  "xyzyx" <xyzyx at inbox.lv>
Subject:  Re:  ux2_bootstrap/Perl build results

On Mon, 22 Jul 2002 22:28:40 -0500 (CDT), Maynard wrote:

>On Mon, 22 Jul 2002 21:01:24 -0500 (CDT), xyzyx wrote:
>
>>Failed 6/726 test scripts, 99.17% okay. 36/68687 subtests failed,
>
>previously reported results in this neighborhood:
>
>Failed 4/726 test scripts, 99.45% okay. 10/68689
>Failed 3/726 test scripts, 99.59% okay.  5/68681
>Failed 4/726 test scripts, 99.45% okay.  8/68656
>Failed 6/726 test scripts, 99.17% okay. 10/68642
>Failed 3/726 test scripts, 99.59% okay.  4/68471

Guess I'm in last place then!

Lots of the errors that scrolled by looked like chmod errors ("unable
to set read+write permissions" or something).

Thanks,
Paul