summaryrefslogtreecommitdiffstats
path: root/solenv
Commit message (Collapse)AuthorAgeFilesLines
* Towards a Mac OS X PythonTestStephan Bergmann2013-04-265-23/+44
| | | | | | | | | | | | | Use URE_BOOTSTRAP instead of listing ~its content individually. [DY]LD_LIBRARY_PATH needs to be set to all of installdir's ure and program as well as OUTDIR, as initPoniesMode (pyuno/source/module/pyuno_module.cxx) tries to load the OUTDIR test library. Abstracted as gb_PythonTest_PRECOMMAND. Abstracted Mac's special LibreOffice.app/Content path as gb_DEVINSTALLROOT, adapted a number of places accordingly. Change-Id: I1646615c77dfab342675f84dbb83c579bf16058d
* Binding OOO layer libs to NONE layer ones is bogusStephan Bergmann2013-04-261-1/+0
| | | | Change-Id: I7470ab98cf68e1fd426983ce185c8fd2d73003e3
* Bad "NONE" layer path prefixStephan Bergmann2013-04-261-1/+1
| | | | Change-Id: I8f6859e172d272887166203919c5a4538c61b6cc
* gbuild: MSVC: only re-link libraries if exported symbols changeMichael Stahl2013-04-261-0/+8
| | | | | | | | | | | Implement commit 3213cd54b76bc80a6f0516aac75a48ff3b2ad67c for MSVC. Not sure what the extra "Summary" section at the end of the output does, and whether it would make sense to filter it. Change-Id: I874abec0e8bb8d1ec2894fd6e94bc78b48e384e1 Reviewed-on: https://gerrit.libreoffice.org/3612 Reviewed-by: Tor Lillqvist <tml@iki.fi> Tested-by: Tor Lillqvist <tml@iki.fi>
* rebuild scp if autoinstalled libs changeDavid Tardon2013-04-252-7/+8
| | | | Change-Id: I76cca4646bd0258cdf64a079edb017e6243839af
* change warning to errorDavid Tardon2013-04-251-1/+1
| | | | | | | It seems that the nonexistent include paths introduced by the headers move have been cleaned up. We do not want to introduce new ones. Change-Id: Ife53bbf0c63e12a41f6b447357472ad41eed5cbe
* fixes for libmergedPeter Foley2013-04-241-1/+4
| | | | Change-Id: I77c7ebcbba4fd2c78d66d6088bbbf4a3122f7c8a
* libpython.so is now in instdirPeter Foley2013-04-241-1/+1
| | | | Change-Id: Ib1b1a853d265a28bfab19c0fa25b778fcec92ba6
* gbuild: AllLangResTarget: add missing dependency on concat-depsMichael Stahl2013-04-241-1/+2
| | | | | | ... was missing from 8379fceb2e7c33e5e0555295be06fde60a6bb240 Change-Id: I664cfabdaf8834b1023dcc7e2a28b584abf10908
* removed cruft, added instdir to filterEike Rathke2013-04-241-2/+2
| | | | Change-Id: I30b1b923658a851535f309c3773f5cbfa697d1bf
* Prevent gb_Helper_PHONY from leaking into the shellTor Lillqvist2013-04-241-1/+1
| | | | Change-Id: Ifec0ea4dd06c9c06a796f293f0fc0973296fed60
* Let climaker work on *.rdb.oldformat for nowStephan Bergmann2013-04-241-4/+4
| | | | | | ...as it still reads type data via com.sun.star.registry.SimpleRegistry Change-Id: Ib65848eb77d220b83f5e70cb98ecf3cb9c46d7de
* gbuild: AllLangResTarget: use concat-depsMichael Stahl2013-04-241-1/+2
| | | | Change-Id: Ieb5e380b6e7a1c097d403f962b4e82b84cce40b5
* gbuild: get rid of processdeps.awkMichael Stahl2013-04-243-94/+9
| | | | | | | | | | | | | | | | | | The only thing that processdeps.awk does that is actually useful is to be able to set the target name freely, but that is only important for the default resource file which was just renamed. Instead hack makedepend to provide the required functionality directly: - write dummy rules for included files so builds don't break on rename/removal - format deps one per line like concat-deps expects - concat-deps expects standard POSIX line ends, so open in binary mode (also disable that awful backup copying nonsense) Though in retrospect it would perhaps be a better investment of time to try to replace makedepend... Change-Id: I54fafdcdcf1a52692e62b8f1f4b96fb9a93d4421
* rename "shlinfo.rc" to "default.rc"Michael Stahl2013-04-242-2/+2
| | | | | | makedepend annoyingly picks up the target file name from the source file Change-Id: I4e7378e3f428dbae42fb99cb1c21ed840c20da40
* gbuild: do not use ' quotes when echoing dependenciesMichael Stahl2013-04-243-4/+4
| | | | | | | | In AllLangResTarget the gb_Helper_abbreviate_dirs replaces the paths in the echo command with $W, which the shell then does not expand due to ' quotes. Mixing gb_Helper_abbreviate_dirs and ' is unsafe. Change-Id: Ia99d185917a55486d5fd67f82d5b323a05e3f5bd
* Generate new format type rdb filesStephan Bergmann2013-04-243-5/+19
| | | | | | | | | | | | | | | | | | | | | | With the main components (theTypeDescriptionManager, codemakers) supporting the new unoidl format, it is time to switch the generated type rdb files to that new format. For now, this is done as a two-stage process, though: Every $(WORKDIR)/UnoApiTarget/*.rdb file in the new format is accompanied by an *.rdb.oldformat file; both are generated in the same [UNO] gbuild step. This is for several reasons: * idlc still generates old format output. * The regcompare step (comparing against a reference rdb to catch incompatible changes) is still based on the old format. (The reg2unoidl tool needs any dependend rdbs as additional inputs, that's why the use_api information needs to be tracked as UNOAPI_DEPRDBS for now. That can be removed again when reg2unoidl is no longer used.) Change-Id: Id625c88f9ecdbaba9e2af7410417e00b4ba36acc
* install gen. IDL docs using filelistDavid Tardon2013-04-241-0/+61
| | | | | | | Change-Id: I474f6e7a45d09683eb0dd7172114407c9dca84d7 Reviewed-on: https://gerrit.libreoffice.org/3571 Reviewed-by: David Tardon <dtardon@redhat.com> Tested-by: David Tardon <dtardon@redhat.com>
* add another filename that shouln't be matched as errorDavid Tardon2013-04-241-0/+1
| | | | | | | | | | This procedure must surely win the prize for the lamest error handling mechanism ever conceived. Change-Id: I90d2a1160ff3e1e870e9ca3f03c92436398c3b7f Reviewed-on: https://gerrit.libreoffice.org/3595 Reviewed-by: David Tardon <dtardon@redhat.com> Tested-by: David Tardon <dtardon@redhat.com>
* Add license filtering to configure.Michael Meeks2013-04-241-1/+1
| | | | Change-Id: Icd53c6ccfcca25d828d79ad62d1e075869998850
* Clean up of dead code, now that UnoApiMerge.mk is goneStephan Bergmann2013-04-241-26/+2
| | | | Change-Id: Ib65496c3bd26a55a713717776184441a959e5469
* allow out-of-srctree builds on windowsChristian Lohmaier2013-04-241-2/+2
| | | | | | | | | | | | config_host (config_$(gb_Side) dir was included from SRCDIR, but when runing the build in another directory, it is in BUILDDIR WinResTarget.mk was forgotten when the other references were updated in ae4e327739112ba326a2945fc0d5550739f3083d and follow-up ones. Change-Id: I8c5a8edb4a17dd33abdcdb04c3af830cad44e07a Reviewed-on: https://gerrit.libreoffice.org/3593 Reviewed-by: Fridrich Strba <fridrich@documentfoundation.org> Tested-by: Fridrich Strba <fridrich@documentfoundation.org>
* UnoApiMerge has become unused nowStephan Bergmann2013-04-244-66/+0
| | | | Change-Id: I9e4e153e1a9ac82d4015b73ecf3a9a2429ab346f
* Get rid of UnoApiMerge_uretypes, which is just a duplicate of UnoApi_udkapiStephan Bergmann2013-04-241-1/+1
| | | | | | | | | What is a little confusing is that the udkapi.rdb ends up as types.rdb in the installation set (in the URE's sub-tree). So all places that reference it during the build do so as "udkapi" while all places that reference it in an installation set do so as "types." Change-Id: I35d0695966b3bd703f5494b636b9782efc0d3fcb
* Update location of log-areas.doxTor Lillqvist2013-04-241-1/+1
| | | | Change-Id: If54a3d7047f13ae9c9345c21737a89afee645403
* make sure the dir existsDavid Tardon2013-04-241-1/+1
| | | | | | | | I am hitting this problem with libcppunit.so.exports. It happens because libcppunit.so is not built by LinkTarget, so nothing has created $(WORKDIR)/LinkTarget/Library/ yet. Change-Id: Ibdde83c746f02110aab113491a2bf37bf2d417eb
* gbuild: fix generated deps for API headersDavid Tardon2013-04-241-0/+11
| | | | | | | | | | This should be reverted after some time, as it solves a one-time problem. Change-Id: Iaa4bc17267a24bd744573e56108b8d207936629c Reviewed-on: https://gerrit.libreoffice.org/3506 Reviewed-by: David Tardon <dtardon@redhat.com> Tested-by: David Tardon <dtardon@redhat.com>
* gbuild: do not deliver UNO API headersDavid Tardon2013-04-242-8/+3
| | | | | | | | | | There is no reason to copy 1000s of files to $(OUTDIR)/inc when we can easily include them from $(WORKDIR). Change-Id: I133ab6506ba76af132143b982f056e652c09d8f5 Reviewed-on: https://gerrit.libreoffice.org/3505 Reviewed-by: David Tardon <dtardon@redhat.com> Tested-by: David Tardon <dtardon@redhat.com>
* move URE headers to include/David Tardon2013-04-241-1/+1
| | | | | | | Change-Id: Ib48a12e902f2311c295b2007f08f44dee28f431d Reviewed-on: https://gerrit.libreoffice.org/3499 Reviewed-by: David Tardon <dtardon@redhat.com> Tested-by: David Tardon <dtardon@redhat.com>
* kill RTLIBS group; and autoinstall the librariesMatúš Kukan2013-04-238-16/+1
| | | | | | This commit changes names for comphelper i18nlangtag and ucbhelper. Change-Id: I93731ef31f894854f9c642d3adee41006552e303
* urelibs: do not merge purpenvhelper (unless mergelibs=all)Matúš Kukan2013-04-231-2/+4
| | | | Change-Id: I147d05b91d46f05461f0548000c891f89486bcc7
* always build also cppu and cppuhelper librariesMatúš Kukan2013-04-231-1/+4
| | | | | | Otherwise auxtargets for these libraries are missing for URELIBS=TRUE Change-Id: Id07b052813609eac0cb68a651f1f976a88168c65
* urelibs: add more librariesMatúš Kukan2013-04-231-0/+4
| | | | Change-Id: Ibe1e811c2ee1db93200b63023dc5922cf1ecb6ca
* remove executable bitLuboš Luňák2013-04-221-0/+0
| | | | Change-Id: I4582e96d97f40b1f592882de4e84036b017d8679
* fix shell script execute permissionsNoel Grandin2013-04-221-0/+0
| | | | | | | | looks like it was inadvertantly removed by commit 62badf38283ef31e50d40e62f5d8bc3824c615d7 "Move to MPLv2 license headers, with ESC decision and author's permission" Change-Id: Iac1d010cf6f4b03e0e7e14e26552f7be31aafb79
* replace python-core zip built in pyuno with direct use of PackageMichael Stahl2013-04-224-4/+4
| | | | | | | | | | | | | | | | | | | - python3: deliver files to INSTDIR, with same layout as instset and do not deliver .lib files - pyuno: remove obsolete python.bin targets - pyuno: remove usage of CustomTarget_zip for WNT and non-Mac UNX platforms (sadly it is apparently still needed for "system" python on MinGW) - scp2: use the python3 filelist There is still a problem here because the installer does not currently allow to preserve the executable bit on files in a filelist - RepositoryExternal: run python executable from INSTDIR and link against libraries in UnpackedTarball dir Change-Id: I931ca0a8be6ff40051b1ca50da1f0770e6057832 Reviewed-on: https://gerrit.libreoffice.org/3525 Tested-by: LibreOffice gerrit bot <gerrit@libreoffice.org> Reviewed-by: Michael Stahl <mstahl@redhat.com>
* gbuild: LinkTarget: try to fix gb_LinkTarget_add_auxtarget for make 3.81Michael Stahl2013-04-221-1/+7
| | | | | | | | Apparently make 3.81 has trouble with the wildcard in there, it never seems to find the file... implement the test in shell instead, since we invoke a command on both branches it shouldn't be slower. Change-Id: I93cc2cf09ae8af1d9379fde314a44ef92a28b434
* add a check to catch typos in Package namesDavid Tardon2013-04-221-0/+15
| | | | | | | | Change-Id: Icae1d4f9f96d8e79e4992f546aad0630f5ae9236 Reviewed-on: https://gerrit.libreoffice.org/3542 Tested-by: LibreOffice gerrit bot <gerrit@libreoffice.org> Reviewed-by: David Tardon <dtardon@redhat.com> Tested-by: David Tardon <dtardon@redhat.com>
* chmod +xTor Lillqvist2013-04-221-0/+0
| | | | Change-Id: I2358b2241bf260afb85005c247cb8be82104e2e5
* Move to MPLv2 license headers, with ESC decision and author's permission.Michael Meeks2013-04-2247-1027/+188
|
* gbuild: work around problems with aux-targets and DLL renames (MSVC)Michael Stahl2013-04-211-1/+5
| | | | | | | | | | When renaming a DLL but not the import lib it can happen that the "touch" rule creates the DLL file which is not helpful. Since there is no easy way to fix this properly, and DLLs are not renamed often, put in a lame hack to detect the situtation and delete the import lib, which will get it re-linked on the next make invocation. Change-Id: I3909cf7d377433d263e517719cca07005bade512
* gbuild: fix spurious delivering of import libs (MSVC)Michael Stahl2013-04-211-2/+2
| | | | | | | | | The DLL aux-target is always newer than the work-dir import lib, which causes spurious re-delivery of the solver import lib. Touch the DLL file to the same timestamp as the import lib to prevent this. Change-Id: I8394201c7a672a50db3af51b9442f8992cdef1d1
* Some changes in qtz handlingZolnai Tamás2013-04-213-58/+110
| | | | | | | | | Executalbes, which work one language, generat qtz by own. (stringex,helpex,treex,propex) So these executables can generate qtz without po file when use them with qtz, call them with "-m" flag without parameter. Change-Id: I56c34db7151dc3ef0ce1c85ed607719e4cbb5e92
* accidentally renamed variableMichael Stahl2013-04-211-1/+1
| | | | Change-Id: I940a807a6bd92393fa4bf01790c1478234228e5e
* gbuild: PythonTest: automatic GDB backtrace on failuresMichael Stahl2013-04-201-12/+19
| | | | | | | | | | | This requires changing the interface a bit to pass the directory containing the test modules separately, because strangely the "unittest" thingy cannot load the test modules by their absolute path when changing the current working dir to where the core dump should happen. So put the directory onto the PYTHONPATH to work around that; arguably it's even nicer for clients this way. Change-Id: I0d1d39f33d0c086c6c3014d2f9ba93f0a3dc0128
* gbuild: PythonTest: try to make it more portableMichael Stahl2013-04-205-5/+17
| | | | | | Mac OS X is not tested though. Change-Id: I143e20e91378b5c0546c0435dda756682d7773e7
* warn if using nonexist. dirs in include pathDavid Tardon2013-04-201-0/+8
| | | | Change-Id: If3c44175a20b7c1720c7200c9bb06372740f95fe
* fdo#63591 pack images referenced by .ui filesDavid Tardon2013-04-193-2/+97
| | | | Change-Id: Ia7e932a98d3f2c8bca981261406798760b69f8e8
* gbuild: fix deps for UIConfig packagesDavid Tardon2013-04-191-2/+3
| | | | Change-Id: I18c270cbcde82c315fd7cf1ba8dfc37ebe1a81aa
* gbuild: PythonTest: do not invoke the python.bin directlyMichael Stahl2013-04-191-1/+1
| | | | | | | | | | | | | This works to run the test (with the fixed RPATH in python.bin from previous commit), but debugging with gdb becomes impossible. The tests need PYTHONPATH set to find the internal python libraries, but gdb also uses PYTHONPATH and then tries to load the libraries there which do not necessarily match the Python version that gdb is built against. Invoking the python wrapper shell script avoids this, and since that "exec"'s python.bin debugging with gdb works too. Change-Id: I6ea2feb44cebeda1ba2c397baa3837f9f1259a08