Commit Graph

21 Commits

Author SHA1 Message Date
2b5dd63565 python3: 15833.patch makes sense with ro /usr on Linux as well
Change-Id: If0d7b17b97a78eddcdd02b3951afb7b2a1ae43ad
2013-02-10 21:42:14 +01:00
eb96e43252 python3: bug 15833
That python bug cause problems when libreoffice is on a read-only
media... which is very common for Mac as the dmg used to package
the produce is seens as a read only volume.
This patch the bug 15833 for MacOSX only since that is the platform that
is most likely to be impacted, and because of bug 15431 that make
patching on Windows more complex/dangerous.

Change-Id: Ie7406c1c75748d38c871b3b544560caa62e9d838
Reviewed-on: https://gerrit.libreoffice.org/1934
Reviewed-by: Norbert Thiebaud <nthiebaud@gmail.com>
Tested-by: Norbert Thiebaud <nthiebaud@gmail.com>
2013-01-31 10:42:45 +00:00
0f6f38b5d7 Fix 64-bit OS X build: don't try any universalness 2013-01-02 20:55:12 +02:00
9ddba66440 convert openssl to gbuild and add to tail_build
Change-Id: I52c62a91e317f072237cf25ed54f3cc6456d82b3
Reviewed-on: https://gerrit.libreoffice.org/1495
Reviewed-by: Peter Foley <pefoley2@verizon.net>
Tested-by: Peter Foley <pefoley2@verizon.net>
2012-12-31 20:13:20 +00:00
4ea11e31e9 /p:VisualStudioVersion=11.0 here, too
Change-Id: Ifa463327e9f33696012b3add0640b12f6d585178
2012-12-18 16:53:42 +02:00
aaee2f77bd Revert "fix python3 build on SLED11"
Failures to build some python modules are actually not fatal, I just got
confused because the whole python build runs in parallel to the normal make.
This reverts commit f4ae375c00deb2297074c59b62db87de080bf591.
2012-12-04 11:53:11 +01:00
f4ae375c00 fix python3 build on SLED11
Apparently all recent systems use ncursesw, for which there is
-I/usr/include/ncursesw, but SLED11 uses ncurses lib, and there's no -I for that.

Change-Id: I61ec795aae45e1074075351eca62299784d08b09
2012-12-04 01:04:53 +01:00
773df2cb86 Nah, wrong to use /p:PlatformToolset=Windows7.1SDK
It must be my local installation of VS2010 that is somehow screwed up
when building here it doesn't find <windows.h>. I need to fix that
instead.

Change-Id: I37a5f8b41f193b108f33464a6a127c0a5969d232
2012-11-30 10:58:47 +02:00
1ea411fb15 We need to tell the MSVS 2010 build to use Windows7.1SDK
At least for me it wouldn't build otherwise. But yeah, what it
somebody uses MSVS 2010 with another SDK? It seems that the solution
only offers the SDK 7.1 as an alternative?

The default was v100, whatever that measn. Could it be that my MSVS
2010 installation is borked? Or that I did not have to install a
bundled SDK with it, because I already had a separate 7.1 SDK?

Also simplify a bit, no need to $(filter) on VCVER inside ifeqs that
already check the very same VCVER.

Change-Id: Ifef98c9466fc24db27d9e38c6878c77adfb4ed75
2012-11-30 01:22:22 +02:00
38fed61f63 Kill the ProjectReference to ssl.vcxproj, too
Otherwise it would try to build the ssl.vcxproj which we don't want
(because we want to use the openSSL already built from solver), and
which fails anyway because for some reason it wants to run
python_d.exe.

Change-Id: I7471bc26ae96be84b976ba35bb959d75678df980
2012-11-30 01:21:09 +02:00
be1136a3f1 python3: use version variables instead of hardcoded number
Change-Id: Ic91cab680b86d8064212e9833c81b37db2002720
2012-11-28 23:49:26 +01:00
d860b4f451 defuzz patches to squeak by RHEL-5 patch
Change-Id: Iac990e65e3af852a527e67154c66e8ad39ce4767
2012-11-28 09:18:09 +00:00
11e776023c python3: try to fix clang breakage in libffi
Thanks kendy for digging up the patch.

Change-Id: I97bc96081736596e84206b95a8d6b658ec3ffae5
2012-11-27 16:57:20 +01:00
ee352d11aa python3: fix MSVC "implicit int" build breaker on tinderbox
Thanks to shm_get for guessing the cause of the problem.

Change-Id: Ieca7199c0c267dc2acaa9ece3ef55747e6a4f816
2012-11-27 16:57:20 +01:00
9d921baa17 clean up PYTHON related version etc. variables:
- configure defines PYTHON_VERSION, PYTHON_VERSION_MAJOR,
  PYTHON_VERSION_MINOR
- remove pyversion.Makefile

Change-Id: I19ac8df18a520ad56bf63ea038dc0769b8249d0b
2012-11-27 16:57:19 +01:00
49313b0626 Make python3 work with custom VALGRIND_CFLAGS
Change-Id: Ia4b08a1b20bf46af4d06c0478ed8e795ee543703
2012-11-27 15:35:02 +01:00
bee01c825b python3: build LibreOfficePython.framework on MacOS X
Change-Id: I0815aa0f5b50166f626f721be56969c0afd655a8
2012-11-26 23:14:34 +01:00
1f8d5eed2a python3: fix typos in makefile
Change-Id: I61ea54ff5a5771ad2dee1b3514c97fbdd9f241b9
2012-11-19 13:37:28 +01:00
f36b2f4320 Using --with-system-expat does seem to work also for a "bundled" one
Change-Id: Iff8904ac0c856dd3175b429b4919a04a57c1b6ad
2012-11-19 12:17:51 +02:00
d3e0fe8213 Make building python3 with current MacOSX tools proceed a bit further
I used the latest Xcode and the 10.7 SDK. Configures, and compiles
(for a while? all that is expected?), but then fails with cp:
.../workdir/unxmacxi.pro/UnpackedTarball/python3/python is a directory
(not copied). That is from ExternalPackage_python3.mk.

No idea how well, if at all, it configures and builds using the Xcode
2 or 3 compiler and 10.4 or 10.5 SDK.

Change-Id: I3ae838263a4db1b67e7c835e567540fac60b98bf
2012-11-17 14:37:02 +02:00
8a6c5b2fcb python3: add module for internal Python 3 build (not active yet)
The module builds here on Fedora 17 and with MSVC2008.
MacOS X is unfinished and probably breaks, which is why the module
is disabled now.

These patches from module python were dropped:

Integrated upstream:
- Python.mipsel-py4305.patch
- Python-2.6.1-py4768.patch
- Python-2.6.1-py2422.patch (modified, use --with-valgrind)
- Python-2.6.1-urllib.patch
- Python-2.6.1-py8067.patch

Obsolete:
- Python-2.6.1-svn-1.7.patch (migrated to non-toy HG now)
- Python-parallel-make.patch
- Python-2.6.1-nohardlink.patch (no idea why that would be needed,
                                 NFS should support hard links)
- Python-2.6.1-sysbase.patch (Solaris 11 setsolar specific patch)
- Python-2.6.1-cross.berkeleydb.patch (berekeleydb removal)
- Python-2.6.2-bdb48.patch
- Python-2.6.1-vc10.patch (upstream supports vc10)

An attempt to cross compile with mingw that proved unsucessful according
to dtardon; there is upstream work on this topic that is possibly
already in 3.3: http://bugs.python.org/issue8067
- Python-2.6.2-cross.patch
- Python-2.6.2-cross.fix-configure.patch

Change-Id: Iba9a3cab955983e173e12110f93a6f381d86f9ce
2012-11-17 00:45:13 +01:00