Additional:

15. I would like to move dmake to head after ensuring the last patch  
for solaris is merged to head.

16. connectivity is  not bison 1.75 compliant, see
http://www.openoffice.org/servlets/ReadMsg?msgId=547579&listName=dev

Resolved:
config_office  is updated.
nas is update - version 1.6
stlport is updated.  New separate bug report not show stopper.

Requires verification:
I have tagged tools/bootstrp files with 643C versions.  To be verified
/ moved by appropriate people.

Here is an interesting problem. unxlngi4.mk the CFLAGS has -g
specified.  This blew my build space of 3.7 gig away.  This should
probably removed.  It is an option to build with -g another way.

Additional:

15a. Use branch vq_mingw for dmake. It is HEAD plus a few modifications
  to make HEAD usable with the "normal" dmake build for MSVC, with
  the batch file. It also holds the modifications to build dmake
  with MinGW (cygwin environment)
  The current cws_srx644_ooo20030223/HEAD version doesn't build for
  w32/4nt

17. ICE in MSVC, see IZ #11261

18. RasDialW problem, see IZ #10727

19. Non-fatal error in readlicense_oo, the target dir doesn't
  exist, I'm not sure if the following readlicense_oo.diff is the
  right place to fix it.

20. w9xucwrp.dll doesn't get build. See IZ #10740, the following
  sal_build.lst.diff fixes this. I got no approval so far.

21. See 16. , fix conn_fix.diff is following

22. Project setup2 needs sysui as dependency, setup2.diff

23. # two icons are missing in desktop/util, not build in sysui:
"cp sysui/win32/office/icons/018_configuration.ico
sysui/oowin32/office/icons/"
"cp sysui/win32/office/icons/019_open_document.ico
sysui/oowin32/office/icons/"
  Plus the following sysui.diff.

  This is a workaround, not a FIX.

24. Change the 001_star_main.ico icon to 001_star_butterfly.ico
  in desktop/util/makefile.mk. See desktop.diff

... the build is continuing

Created attachment 4611 [details]
Patch for readlicense_oo

Created attachment 4612 [details]
Patch for sal

Created attachment 4613 [details]
Patch for connectivity

Created attachment 4614 [details]
Patch for setup2

Created attachment 4615 [details]
Patch for sysui

Created attachment 4616 [details]
Workaround for desktop

I forgot:

25. Define NO_HIDS=TRUE, see:
  http://www.openoffice.org/issues/show_bug.cgi?id=11056

Btw: The build just finished :-) but a lot of files are missing in
  the last instsetoo stage plus:

WARNING! Project(s):
rvpapi
sane
helpcontent

not found and couldn't be built. Correct build.lsts.

Hi,      
      
readlicense_oo.diff approved      
    
sal_build.lst patch (looks good but I can't test it - approved)     
    
connectivity patch - wasn't something like this approved for 643    
                     and 102 - so should be okay for this tree    
   
setup2.diff - approved   
  
sysui.diff and movement of icons to oowin32 - approved  
            (this was my fault not to keep sysui/oowin32 up to  
             date when things changed in sysui/win32)  
  
 
I would like someone from desktop to approve your desktop diff since 
the whole icon issue there (see blue star versus ooo in point 13 of 
this issue). 
 
As to your point 25: 
 
- rvpapi is listed in vcl as a dependency but does not exist so we 
can simply remove it from the vcl/prj/build.lst 
 
- helpcontent is missing (see point 7 above) 
 
- you can check out sane as an independent module but it is 
  not built as part of the tree (it just delivers a sane header). 
 
 
Kevin 
 
 
 
  
  
    
    

Hi, 
 
I just commited a fix for product/util to handle salhelper and 
cppuhelper naming differences across UNX and WNT. 
 
I also added "filter" and "psprint_config" to the 
instsetoo/prj/build.lst tsince they won't get built at all 
otherwise and they both are needed. 
 
So Looking at our list,  the following points have now been 
resolved: 
 
points: 2, 3, 4, 5, 6, and 8 
 
Ken what is the status of dmake?  The "best" dmake we have in a 
released tree is in OOO_STABLE_1 (it is the same as OO643C but with 
the Solaris fixes in place). 
 
Has anything been tagged to replace this dmake so something works? 
 
Thanks, 
 
Kevin 
 

readlicense_oo.diff committed
sal_build.lst.diff committed
setup2.diff committed
sysui.diff and 2 copied icons to oowin32 - committed

I'm waiting for a third opinion on conn_fix.diff ;-)

Hi Ken,   
   
I am trying to track down points 10 and 11 where prj/d.lst files   
seem to be corrupt.   
   
It seems there has been some changes in solenv/bin/deliver.pl that   
are related to this. These two prj/d.lst files use a new macro   
COMMON_DEST which does not seem to get set or set properly.   
   
Here is an example of the line in extras/prj/d.lst   
   
..\source\truetype\standard_sans\*.ttf %COMMON_DEST%\pck%_EXT%\*.ttf   
 
I believe that non-platform specific files are delivered to some   
non-platform specific "common" solver areas under Hamburg's build   
environment.   
   
If you look in solenv/bin/deliver.pl you see the following changes:   
(just a small sample) 
   
  @@ -111,11 +111,13 @@ 
 $dlst_file          = 0;            # path to d.lst 
 $umask              = 22;           # default file/directory 
creation mask 
 $dest               = 0;            # optional destination path 
+$common_dest        = 0;            # common tree on solver 
 
 @action_data        = ();           # LoL with all action data 
 @macros             = ();           # d.lst macros 
 @hedabu_list        = ();           # files which have to be 
filtered through hedabu 
 @zip_list           = ();           # files which have to be zipped 
+@common_zip_list    = ();           # common files which have to be 
zipped 
 
 $files_copied       = 0;            # statistics 
 $files_unchanged    = 0;            # statistics 
@@ -167,10 +169,13 @@ 
     glob_and_copy($from, $to, $touch); 
 
     $line =~ s/%__SRC%/%COMMON_OUTDIR%/ig; 
-    $common = expand_macros($line); 
-    ($from, $to) = split(' ', $common); 
-    print "copy common: from: $from, to: $to\n" if $is_debug; 
-    glob_and_copy($from, $to, $touch); 
+    if ( $line =~ /%COMMON_OUTDIR%/ ) { 
+        $line =~ s/%_DEST%/%COMMON_DEST%/ig; 
+        $common = expand_macros($line); 
+        ($from, $to) = split(' ', $common); 
+        print "copy common: from: $from, to: $to\n" if $is_debug; 
+        glob_and_copy($from, $to, $touch); 
+    } 
 } 
 
 
I do not think $common_dest is ever being properly set to anything. 
 
So we seem to have a problem here. 
 
Kevin 
 
 

Hi, 
 
I am sure Martin does not want to see all of the reminders from this 
issue!!! 
 
So I added Marting as CC and changed ownership to myself (the 
porting co-lead) since I ignore all those reminders anyway!) 
 
;-) 
 
Kevin 
  

Hi Volker, 
 
The  connectivity/source/parse/sqlbison.y	 piece of the patch you 
submitted has already been approved for ooo102 since I remember 
committing those fixes at some point. 
 
That only leaves the minor makefile and include issues. 
 
I will rebuild connectivity with your patch to verify they are not 
an issue and if all is okay I will approve it then. 
 
Thanks, 
 
Kevin 
 

Created attachment 4623 [details]
deliver.pl fix common_outdir set to null string -> wrong common_dest

Hi, 
 
I just added a patch for deliver.pl that will fix the issue 
of how COMMON_DEST is used in extras/prj/d.lst and 
psprint_config/prj/d.lst 
 
It seems we set COOMON_OUTDIR ($common_outdir) to ""  (see 
config_office) but given the changes in deliver.pl this resulted in 
COMMON_DEST ($common_dest) being incorrectly set to just 
$solarversion. 
 
The patch I created tests for this case and properly sets 
$common_dest to the right place when $common_outdir is the null 
string. 
 
Can someone test and approve this change? 
(i.e. do the following: 
 
cd extras 
deliver -force 
 
See where and what is copied 
 
Then apply the patch to deliver.pl 
 
and repeat the  
 
deliver -force 
 
and now the proper files and destinations should be given. 
 
Thanks, 
 
Kevin 
  

Hi Volker, 
 
I just finished testing your conn_fix.diff patch and all went well 
(under the old bison). 
 
Since the key element of the patch was already approved for 
OOO_STABLE_1 and since it applies and builds cleanly even on old 
bison, I approve the patch. 
 
Kevin 
 

conn_fix.diff - committed

Ooop, after committing the patch I saw that my patch program also
fixed some ^M lineendings in sqlflex.l. 125 changed lines, but
only two visible changes. Hope this doesn't hurt :-(

http://tools.openoffice.org/source/browse/dba/connectivity/source/parse/sqlflex.l.diff?r1=1.15.12.1&r2=1.15.12.1.24.1

Just for your information. I have committed all Linux for S/390
related patches. The issues are:

8385
8387
8388
10150
10152
10153

Gerhard

Hi Gerhard, 
 
Are all S390 patches now in place?  Are there any outstanding issues 
for S390 that have to be approved yet or are we good to go? 
 
Thanks, 
 
Kevin 
 

Hi Kevin,
everything is committed now for Linux for S/390.

Gerhard

Hi,  
 
That's great!  So S390 is done!  Have you been able to get 
any 644 build completed yet? 
 
Also Remaining ports to integrate patches for 
 - MacOSX 
 - FreeBSD 
 - Irix  
 - NetBSD 
 
Are there any other active ports out there yet that might have 
patches we need to integrate? 
 
Kevin 
 

Added some dependencies. Issues: 10855 (solved), 11261 (17.),
10727 (18.), 11056 (25.)

I cannot commit any time this week, on a course however...

dmake on HEAD I have built on Mac (thanks to mac person) and the
linux.  The solaris patch theat you made I asked to be merged to head,
we need to check posix.h

There is the patch that VOlker made for tempnam also that needs merging.

The dmake on head has all the warnings removed and there are some
errors in the code that have been removed.

Second try. Does issuezilla have Alzheimer's disease?
The dependencies were there and vanished after Kens entry.

Added some dependencies. Issues: 10855 (solved), 11261 (17.),
10727 (18.), 11056 (25.)

plus MSVC dmake issue 10910.

Hi,

I had a look at deliver_fix.diff, looks good and the test shows
that the files in extras now get correctly deliverd to
  solver/644/wntmsci7.pro/...
instead of
  solver/644/.pro/...

deliver_fix.diff - approved

helpcontent has been retagged now.

Created attachment 4632 [details]
patch for ridljar/prj/build.lst to make it build for me.

Hi,   
   
I added the deliver.pl fix to make the proper output directory for   
delivering things under OOo (now committed).  That along with 
Volker, Ken's and Martin's fixes, I think the following points have 
been solved:   
   
2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 16, 19, 22, and 23.  
  
I am unsure about point 20?  Is this now fixed as well?  
  
So that leaves all of the unresolved pending issues attached to this  
one and ...  
  
1, 15, 15a all dmake related - I will grab HEAD of dmake tonight and  
try to integrate in everything we need so that this gets done.  
  
12, 13, 14, 17, 18, 24, and 25 plus the ridljar build order patch 
(we can call that 26!). 
 
So we are getting closer and closer 
 
Kevin 
 

Hi, 
 
I removed the gallery644*.res line as instructed to by Martin and so 
you can add point 12 to the list of resolved items. 
 
Kevin 
 

Hi Kevin,

> I am unsure about point 20?  Is this now fixed as well?  

you gave me approval and I committed the fix. It's working
for me (TM) now.

 Hi, 
 
Ken I approve your ridljar build order patch.  I think this is 
needed for jdk 1.4 only but it can't hurt jdk 1.3.1 and is correct. 
(that takes care of point 26) 
 
Please commit it when you get a moment 
 
Also since 20 is done all that really leaves is 
 
13, 14, 17, 18, 24, and 25 
 
Oliver is reworking sysui to integrate OOo with Star dekstop code so 
14 (CDE 8bit icons display icon corruption) should probably wait 
until that integration is done. 
 
I think 13 (Blue Star) and 24 (butterfly verus ooo) are both issues 
related to desktop integration and can probably wait until that gets 
reworked. 
 
Volker if you can create a HID patch for config_ofice I will review 
and approve it and that will take care of 25. 
 
Then all that will really leave is 17 and 18 which are bing tracked 
under their own issues. 
 
Thanks, 
 
Kevin 
 
 
 
 
 

Hi Kevin.

>Volker if you can create a HID patch for config_ofice I will review 
>and approve it and that will take care of 25. 

I attached the patch to issue 11056.

Hi,

I just closed: 18. RasDialW problem, see IZ #10727

Volker

the good news: I will put rvpapi sources to OOo repository the next
days on this branch.

sane is missing in the CVSROOT/modules alias for all and
OpenOffice.org, I will add this shortly.

New break in build...

gandalf: /data3/office/sysui
$ build
build -- version: 1.79.2.16
 
/data3/office/sysui/oounix/office/scripts
-------------
 
ERROR:
 (/data3/office/sysui/java/office/icons) not found!!
 

One more issue :-(

26. For the W32 build (at least)
   external/ado
           /atl
           /glibc
   Have to be used from OO643C. Please approve this "upgrade".

27. The extensions module needs "twain" as dependency

27a. twain is missing in the OpenOffice module list.

Hi Volker, 
 
But twain is tagged properly, right?  Its just that we have to 
manually check it out as a project.   
 
Martin who do we ask (Louis?) to modify the cvs root alias which 
defines OpenOffice to include "twain"? 
 
I have manually checked out twain and I have added to the extensions 
build.lst in my own tree and will rebuild extensions to test it. 
 
Once it passes (and it will) I will approve it and you can commit 
the change. 
 
Thanks,  
 
Kevin 
 

Hi, 
 
rebuild of extensions with proposed changes after checkout and build 
of twain finished fine. 
 
So I approve the change.  Volker please commit it when you have a 
moment. 
 
Thanks, 
 
Kevin 

I had to apply Ken's patch to get ridljar to build for me.  I have
Blackdown JDK  version 1.3.1.

Hi Chris, 
 
Can you go ahead and commit that fix to ridljar so that it makes it 
into the merge? (I already approved it - see above). 
 
Thanks, 
 
Kevin 
 

twain, sane and rvpapi are now contained in the modules aliases.
rvpapi still only contains prj/d.lst and build.lst, I hope that I can
provide sources tomorrow.

OK, Ken's ridljar patch is checked in:

/cvs/udk/ridljar/prj/build.lst,v  <--  build.lst
new revision: 1.9.12.1; previous revision: 1.9

haggai->mh: (The rvpapi files in prj are only on HEAD so they don't
appear on branch checkouts, but I guess you probably know that :)

Created attachment 4657 [details]
xmlscript needs com/sun/star/xml/sax/SAXException.idl, which comes from offapi project

Created attachment 4658 [details]
from mws_srx644 1.17.4.1->1.17.4.2: #100000#, #103111# As everything inside sysui\java\office\icons has been removed my MH (1.1.30.1), that subdirectory has to be removed here in build.lst.

Hi,
 
>rebuild of extensions with proposed changes after checkout and build 
>of twain finished fine. 
> 
>So I approve the change.  Volker please commit it when you have a 
>moment.

just got back from work, I'll check it in in a minute.

But, please approve also 26. It's WNT specific, it can nearly count
as an obvious fix.
 
Thanks, 
 
   Volker

Hi Volker, 
 
I think you should do a  
 
cvs update -dP in external 
 
The directories atl, glibc, and ado already exist in my build and 
are in the prj/build.lst file. 
 
Also I diffed the external/glibc from ooo643 to ooo644 (my tree 
names) and it showed no diferences.  Ditto for external/atl. 
And there was only a one line change  in external/ado as well. 
 
Perhaps you  checkout external while it was still being updated? 
 
Kevin 
 

Hi Chris, 
 
I approve both the xmlscript build depdencies addition patch and the 
remove the sysui java/icons from the build.lst patch. 
 
Please commit them both when you get a moment. 
 
Thanks! 
 
Kevin 
 

java/icons is now removed from build.lst.

Hi Kevin,
 
>I think you should do a  
> 
>cvs update -dP in external 
> 
>The directories atl, glibc, and ado already exist in my build and 
>are in the prj/build.lst file. 
>..
>Perhaps you  checkout external while it was still being updated? 

Hmm, see what I get when I do:
$ cd external
$ cvs up -dP -rcws_srx644_ooo20030223
$ oosshcvs diff -u -rcws_srx644_ooo20030223 -rOO643C ado atl glibc >
external.diff

This external diff is following as attachment. It's not usefull
for a patch because ado/ado.pat.tar.gz differs and is a binary file.

I *really* need ado, because the location of the header files changed
from: ..\download\AdoCtint.h to %PSDK_HOME\Include\AdoCtint.h
but atl and glibc would be nice because the include the w32-tcsh
patches.

Created attachment 4660 [details]
No real patch, only for comparison.

Fix for 15a. ... The current cws_srx644_ooo20030223 version doesn't
  build for w32/4nt.

The following dmake.diff makes dmake usable for w32-4nt. Please approve.

Created attachment 4664 [details]
Patch for dmake (w32-4nt)

Hi Volker, 
 
I understand now.  My OO643C tree is older than those commits. 
 
So yes, updating them to work with WNT is approved. 
 
(actually these changes must have been previously approved for 
OO643C so you really don't even need my approval)! 
 
Hope this helps, 
 
Kevin 
  

Created attachment 4668 [details]
HAVE_WAIT_H is not defined, use _MSC_VER instead - new patch

external upgrade (26.) - done

27. - done

xmlscript dependency on offapi committed:

/cvs/script/xmlscript/prj/build.lst,v  <--  build.lst
new revision: 1.5.6.1; previous revision: 1.5

Need approval for this psprint patch:

RCS file: /cvs/oo/gsl/psprint/prj/build.lst,v
retrieving revision 1.3
diff -u -b -B -r1.3 build.lst
--- prj/build.lst       2 Aug 2002 12:10:07 -0000       1.3
+++ prj/build.lst       12 Feb 2003 12:02:27 -0000
@@ -1,4 +1,4 @@
-pp     psprint :       tools unotools NULL
+pp     psprint :       tools unotools cpputools NULL
 pp     psprint                                                      
         usr1     -       all     pp_mkout NULL
 pp     psprint\source\fontsubset                               nmake
  -      all      pp_fontsset NULL
 pp     psprint\source\printer                                  nmake
  -      pp_printer NULL

In order to compile under gcc 3.2 the rtti must be removed from the
compile options.  This slightly increases the compiled size.  Please
approve patch.

gandalf: /data2/office/solenv/inc
$ cvs diff unxlngi4.mk
Index: unxlngi4.mk
===================================================================
RCS file: /cvs/tools/solenv/inc/unxlngi4.mk,v
retrieving revision 1.12.12.3
diff -u -b -B -r1.12.12.3 unxlngi4.mk
--- unxlngi4.mk 31 Jan 2003 15:48:17 -0000      1.12.12.3
+++ unxlngi4.mk 12 Feb 2003 13:49:23 -0000
@@ -103,7 +103,7 @@
 CFLAGS_NO_EXCEPTIONS=-fno-exceptions
  
 # -fpermissive should be removed as soon as possible
-CFLAGSCXX= -pipe -mcpu=pentiumpro -fno-for-scope -fpermissive -fno-rtti
+CFLAGSCXX= -pipe -mcpu=pentiumpro -fno-for-scope -fpermissive
  
 # HACK: enable Hamburg developers to build on glibc-2.2 machines but
compile vs. glibc-2.1 headers
 .IF "$(BUILD_SPECIAL)"==""

15a. A similar patch was approved in issue 10910 - committed - done.

That leaves:
  17. ICE in MSVC, see IZ #11261 - patch is there, but not approved.
  and 13./24. the desktop/blue star issue. What's the status there?
for W32.
  

Hi Ken, 
 
perhaps the way to solve this is to ifdef this switch based on the 
gcc version. 
 
So for gcc 3.0.4 you could leave it -fno-rtti 
 
For higher gcc version you remove that (in fact PPC Linux adds 
-frtti for all builds). 
 
At least that way, if Hamburg does not want it, it won't impact 
them. 
 
My 2 cents, 
 
Kevin 
 

Hi Ken, 
 
psprint built fine for me.  What broke for you or what requires 
cpputools to be built first? 
 
Perhaps my build order was just lucky? 
 
Kevin 
 

Hi Kevin, Ken,
I approve Ken's cvs/oo/gsl/psprint/prj/build.lst patch. I need it also
on s390, since psprint/source/helper/helper.cxx includes
cpputools/jenv.hxx now.

Gerhard

psprint has been updated.

I also need an update to a newer version for gcc 2.95 compatibility:
We need to move the tag on comphelper/source/property/propagg.gxx to
version 1.8

Hi,
I am building a checkout from this morning on s390. It looks good so
far, but I need approval for the following additional patch

Thanks,
Gerhard

--- cpputools/source/javaenv/javainfo_impl.cxx.bak      Thu Feb 13
20:39:12 2003+++ cpputools/source/javaenv/javainfo_impl.cxx  Thu Feb
13 14:30:38 2003
@@ -113,6 +113,8 @@
 #define ARCH "ppc"
 #elif defined (MIPS)
 #define ARCH "mips"
+#elif defined (S390)
+#define ARCH "s390"
 #else
 #error unknown plattform
 #endif

Hi Gerhard,

Your patch to cpputools to add the s390 arch is approved.

Please commit it when you get a moment.

Thanks,

Kevin

28. A dmake test in instsetoo/util shows that:

  error: source file 'fop.dll' cannot be found.
  error: source file 'fps.dll' cannot be found.

These files are created in fpicker, and in OO643C fpicker was
in the build.lst, now it is missing.

Ok to add fpicker to instsetoo build.lst ?

Volker

fpicker was missing in instsetoo (28.) - done

Approved by Kevin in dev@porting

Hi!

I had to reopen issue 11019, because it removed some *.bmp which are
obviously needed for W32-* builds.

Point 24. - Change the 001_star_main.ico icon to 001_star_butterfly.ico
  in desktop/util/makefile.mk is still an issue. If you look at the
  makefile you see that the 001_star_main.ico is only needed in WNT
  only sections of the makefile.
  There is *no* 001_star_main.ico in my solver.
  
  I will start an extra issue for this problem, but I think it's to
  late to get this solved before the merge.

Volker

Created issue 11506 for point 24.

setting to fixed, since branch is closed and all deps resolved.

if there is anything left here, we should collect issues in new task
for next cws.

Hi Martin,  
  
I think the only outstanding issues for the next round are:  
  
1. get MacOSX patches in  
  
2. get remaininging FreeBSD patches in  
  
3. fix desktop to use OOo and not Blue Star icons  
  
4. Other dekstop icons issue (Volker's issue)  
  
5. Redesign sysui_oo to properly handle desktop   
   integration so that OOo kde and gnome pieces  
   do not mess up Star ones  
  
6. find a place to put the remaining peices from product/util/  
   makefile.mk since "product" is going away  
  
As for point 5, Oliver has been doing a redesign in sysui_so and he  
has been helping/teaching me how to create something similar for  
sysui_oo so that both modules can sit side by side without causing  
difficulties.  
  
I will open a new issue with these once the new 644 tree reopens for 
commits. 
 
Hope this helps,  
  
Kevin  

Reopened, 001_star_main.ico still missing for WNT

mh-vq: please update sysui for 001_star_main.ico.

vq->mh: Ok, icon gets delivered now.

in cws_srx644_20030223

As mentioned on the qa dev list on March 5th I will close all resolved
<wontfix/duplicate/worksforme/invalid> issues. Please see this posting for details.