Age | Commit message (Collapse) | Author |
|
At least the unrelated
<https://ci.libreoffice.org/job/gerrit_linux_clang_dbgutil/132372/> was killed
after 20 minutes of no stdout/-err activity, with CppunitTest_sc_opencl_test
still running. From its captured state, it looks like it was indeed still
making progress rather than being stuck: During
ScOpenCLTest::testLogicalFormulaXor, the main thread was waiting at
ScFormulaCell::InterpretFormulaGroupThreading ->
comphelper::ThreadPool::waitUntilDone, while four
comphelper::ThreadPool::ThreadWorker threads were each at
> #4 0x00007f70652f9234 in __gnu_debug::_Safe_iterator_base::_M_detach() () at /lib64/libstdc++.so.6
> #5 0x00007f7048d3e8e5 in __gnu_debug::_Safe_iterator_base::~_Safe_iterator_base() (this=0x7f703daa9f20) at /opt/rh/devtoolset-7/root/usr/lib/gcc/x86_64-redhat-linux/7/../../../../include/c++/7/debug/safe_base.h:100
> #6 0x00007f7048d5d709 in __gnu_debug::_Safe_iterator<__gnu_cxx::__normal_iterator<mdds::mtv::base_element_block* const*, std::__cxx1998::vector<mdds::mtv::base_element_block*, std::allocator<mdds::mtv::base_element_block*> > >, std::__debug::vector<mdds::mtv::base_element_block*, std::allocator<mdds::mtv::base_element_block*> > >::~_Safe_iterator() (this=0x7f703daa9f18) at /opt/rh/devtoolset-7/root/usr/lib/gcc/x86_64-redhat-linux/7/../../../../include/c++/7/debug/safe_iterator.h:86
> #7 0x00007f7048d5d6cd in mdds::mtv::soa::detail::iterator_updater<mdds::mtv::soa::multi_type_vector<sc::CellStoreTraits>::const_iterator_trait>::grouped_iterator_type::~grouped_iterator_type() (this=0x7f703daa9ec8) at workdir/UnpackedTarball/mdds/include/mdds/multi_type_vector/soa/./iterator.hpp:74
> #8 0x00007f7048d5d69d in mdds::mtv::soa::detail::iterator_updater<mdds::mtv::soa::multi_type_vector<sc::CellStoreTraits>::const_iterator_trait>::~iterator_updater() (this=0x7f703daa9e98) at workdir/UnpackedTarball/mdds/include/mdds/multi_type_vector/soa/./iterator.hpp:55
> #9 0x00007f7048d5d665 in mdds::mtv::soa::detail::const_iterator_base<mdds::mtv::soa::multi_type_vector<sc::CellStoreTraits>::const_iterator_trait, mdds::mtv::soa::detail::iterator_base<mdds::mtv::soa::multi_type_vector<sc::CellStoreTraits>::iterator_trait> >::~const_iterator_base() (this=0x7f703daa9e98) at workdir/UnpackedTarball/mdds/include/mdds/multi_type_vector/soa/./iterator.hpp:309
> #10 0x00007f704900b9bd in ScValueIterator::GetThis(double&, FormulaError&) (this=0x7f703daaa078, rValue=@0x7f703daaa1f8: -18.819375178914722, rErr=@0x7f703daaa1f6: NONE) at sc/source/core/data/dociter.cxx:141
> #11 0x00007f704900c652 in ScValueIterator::GetNext(double&, FormulaError&) (this=0x7f703daaa078, rValue=@0x7f703daaa1f8: -18.819375178914722, rErr=@0x7f703daaa1f6: NONE) at sc/source/core/data/dociter.cxx:297
> #12 0x00007f70496c3619 in ScInterpreter::ScXor() (this=0x48b93a0) at sc/source/core/tool/interpr1.cxx:1525
with three of them apparently waiting in pthread_mutex_lock but one in
pthread_mutex_unlock (and thus presumably making progress). It looks like with
a debug-mode libstdc++ these worker threads can easily compete for these
listdc++-internal debug-mode mutices. And execution of --enable-dbgutil
CppunitTest_sc_opencl_test on my Linux laptop under load easily took 35 minutes,
so it looks plausible that that unit test was just too big.
While splitting it in two (arbitrarily taking the first and second half of the
CPPUNIT_TEST list), use the opportunity to drop the redundant "_test" from the
two new target names CppunitTest_sc_opencl-1 and CppunitTest_sc_opencl-2.
Change-Id: I9b8f148db667ff9bbf1aacdcaf150e5cb9b8ae87
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/146252
Tested-by: Jenkins
Reviewed-by: Stephan Bergmann <sbergman@redhat.com>
|
|
Change-Id: I53b8564a8fa7ebdc6ec4214a2cecd2917105a777
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/143186
Tested-by: Jenkins
Reviewed-by: Xisco Fauli <xiscofauli@libreoffice.org>
|
|
Some notes:
* testLinkedGraphicRT needed to be adapted. Saving to
Calc Office Open XML doesn't keep IsLinkedGraphic.
Reported in tdf#152036. No idea why it didn't fail before
* Some tests used "generic_HTML" as export filter format, which
is wrong. Instead, "HTML (StarCalc)" should have been used.
But then, the test would have failed with
- unsatisfied query for interface of type com.sun.star.sheet.XSpreadsheetDocument!
So trying to export to html in those tests is not expected.
No idea why it didn't fail before
* testTdf90299 fails now with
- Expected: 'file:///tmp/lu23738863wthph.tmp/dummy.xls'#$Sheet1.A1
- Actual : 'file:///tmp/lu23738863wthph.tmp/lu23738863wthpg.tmp/dummy.xls'#$Sheet1.A1
I added a TODO to check it later
* testNamedRangeBugfdo62729 fails on mac with an infinite loop
Change-Id: I63168583a47f510d1ac22971a274e28d085de61b
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/142576
Tested-by: Xisco Fauli <xiscofauli@libreoffice.org>
Reviewed-by: Xisco Fauli <xiscofauli@libreoffice.org>
|
|
This separates the drawinglayer core functionallity into a
separate library, to keep a strict separation what is backend
dependent and what is not. More strict separation can be done
at a later date.
This will make it possible to push part of drawinglayer
(part of processor2d) directly into VCL.
Change-Id: Ibc26580067e50bf20d7cdd37fa0e44eb10200878
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/127286
Tested-by: Jenkins
Reviewed-by: Noel Grandin <noel.grandin@collabora.co.uk>
|
|
And make it simple to disable the whole feature by setting
gb_DISABLE_PCH_REUSE=1, just in case.
Also work around a possible BOOST_ALL_NO_LIB mismatch when
using the common PCH.
Change-Id: I96fd507edf1ada6242ac225026250e5a588d0193
Reviewed-on: https://gerrit.libreoffice.org/79365
Tested-by: Jenkins
Reviewed-by: Luboš Luňák <l.lunak@collabora.com>
|
|
Generate Application.DocumentChange and Application.Quit events.
SfxHintId::DocChanged seems to correspond nicely enough to
Application.DocumentChange. It is generated a bit eagerly, but as its
documentation is fairly vague and no specific detailed information is
passed in parameters anyway, it probably doesn't hurt if a client gets
it a bit more often with LO than with some other product.
Can now remove the FIXME-marked Quit event things in SwVbaApplication.
Now need oovbaapi in many makefiles for them to compile.
Change-Id: I4d0c5b93b584f198bcc854002eec7aaba7909ecc
Reviewed-on: https://gerrit.libreoffice.org/55106
Reviewed-by: Tor Lillqvist <tml@collabora.com>
Tested-by: Tor Lillqvist <tml@collabora.com>
|
|
This is preparing to change how GraphicManager works where it
won't base itself around GraphicObject anymore but Graphic. No
functional or cosmetic change was made to the classes, only
changes that were needed because of the move and rename.
The only thing that wasn't moved is the GraphicRenderer as it
is not needed in vcl for now (but makes sense to move it in the
future to keep graphic stuff together).
grfmgr was renamed to GraphicObject as the GraphicManager will be
changed a lot and most likely moved out, so the name grfmgr won't
make any sense anymore.
All the UNO implementations were renamed with a prefix Uno and
used the same name as the class name. This is made to be more
specific which are the Uno objects (for example graphic.cxx
contained the implementation of XGraphic, which is similar to
graph.cxx contains Graphic).
Change-Id: I54a2fa6c7e997469aaa7770db05244adb9f64137
Reviewed-on: https://gerrit.libreoffice.org/51068
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Tomaž Vajngerl <quikee@gmail.com>
|
|
Uwinapi is discontinued.
Change-Id: I063b4d0d8fab2d60de168e960a63b8181158ac01
Reviewed-on: https://gerrit.libreoffice.org/23198
Reviewed-by: David Ostrovsky <david@ostrovsky.org>
Tested-by: David Ostrovsky <david@ostrovsky.org>
|
|
Change-Id: I897e5ae6ed0fea89c1f1b9e30bd6cc6505318a17
Reviewed-on: https://gerrit.libreoffice.org/27692
Reviewed-by: Tor Lillqvist <tml@collabora.com>
Tested-by: Tor Lillqvist <tml@collabora.com>
|
|
This patch changes occurences in makefiles
from
$(eval $(call gb_CppunitTest_use_api,comphelper_test_config, \
udkapi \
offapi \
))
to
$(eval $(call gb_CppunitTest_use_sdk_api,comphelper_test_config))
(corrected instead of abandon, jani)
Change-Id: Ic96ec65d82d7452e288f05a8b6d576ef543b068e
Reviewed-on: https://gerrit.libreoffice.org/23426
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: jan iversen <jani@documentfoundation.org>
Tested-by: jan iversen <jani@documentfoundation.org>
|
|
For consistency with what 38cd1d9a5f00dc69ba0d60e2fe6957090c7a26d1 "clew: move
this foreign code to external/clew" did for clew.c. Library_scfilt depends on
clew/clew.h, simply make it depend on the full clew external.
Change-Id: Idcbf56a69084e108832c8970507ebea0f7ff400b
Reviewed-on: https://gerrit.libreoffice.org/24035
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Stephan Bergmann <sbergman@redhat.com>
|
|
This caused incremental builds to fail after today's orcus upgrade.
The cxx files of these tests don't actually include orcus headers, so
the libraries were not re-linked in incremental builds.
Change-Id: Ibfcf216a7848f3c0adfae4ef59c4b756c290d6b2
|
|
... years ago in commit e1b51e7beb7f9cfa7b574b9c2a69799e62963a09
Change-Id: I588370796830dc379f6a004ec7e673b020360eb5
|
|
Change-Id: Ib591b4328589982cda762ce2b1a688345b1665b0
Reviewed-on: https://gerrit.libreoffice.org/13652
Reviewed-by: Caolán McNamara <caolanm@redhat.com>
Tested-by: Caolán McNamara <caolanm@redhat.com>
|
|
There is no obvious authoritative upstream for clew anyway, so it causes
philosophical problems for distros. For a while, we used to use a zip archive
from the "clcc" project on SourceForge that included clew.c and
clew.h. (Before that we also just had clew.c and clew.h in our source repo.)
So, drop the external/clcc module and have clew.c and clew.h in the source
repo again. But this time clew is in a module of its own, not in sc.
This re-introduces "No need to have OpenCL optional at configure-time"
This reverts commit 764836cb00e8e6dfd2ab48e080a166ec90359e01.
Change-Id: I413142f4f9f8399489f9c3e5327132822f07a454
Reviewed-on: https://gerrit.libreoffice.org/13368
Reviewed-by: David Tardon <dtardon@redhat.com>
Tested-by: David Tardon <dtardon@redhat.com>
|
|
As long as this requires bundling clcc, not allowing to disable it is
unacceptable.
This reverts commit 3b0b76462aa54f60bc9823d9c6ef64e5e35c1208.
Conflicts:
configure.ac
Change-Id: I12b3d5e3325bbd03cd98e1fb5139705d694493c2
|
|
Intermediate commit. More changes will follow: The device selection
logic needs to be moved, too. (And cleaned up.) Instead of the
separate formulacalculationoptions dialog we should simply have a
normal options page for those OpenCL-related settings that will remain
purely Calc-specific, like the formula opcode subsetting.
Change-Id: Id60d95e80d377cbbf5780beb473b221bce06b5e5
|
|
Change-Id: Ifb7e86b078bd549506a9cc1b9ce9fc22fffc5eec
|
|
Change-Id: I3a207f86611a5078cd2de1437cd48e612232aaf2
|
|
Just link that code into the sc library. We use clew anyway to handle
optoinal presence of OpenCL at run-time.
This is a straightforward change for now. Could be cleaner; we could
get rid of the fairly pointless wrappers in
sc/source/core/opencl/formulagroupcl.cxx for instance.
Change-Id: I23215c6516ef04698f6891e2a7fc6611dc83b37a
|
|
Change-Id: I38cda4e1ba520c9bb15f71b27543cb80fd696424
|
|
This required some changes to the framework:
* Init-/DeInitVCL is no longer done per individual test in BootstrapFixture, but
once per CppunitTest invocation in a new vclbootstrapprotector (similarly to
the exisiting unobootstrapprotector). CppunitTests that need VCL now need to
declare gb_CppunitTest_use_vcl.
* For things to work properly, the UNO component context needs to be disposed
from within DeInitVCL (cf. Desktop's Application::DeInit called from
DeInitVCL). The easiest solution was to introduce an
Application::setDeInitHook (where the hook is called from DeInitVCL)
specifically for vclbootstrapprotector to call.
* PythonTests don't (yet) call DeInitVCL; they still hook into
BootstrapFixture's original test_init functionality (to call InitVCL), and do
not make use of the vclbootstrapprotector.
Change-Id: I4f3a3c75db30b58c1cd49d81c51db14902ed68b2
|
|
Unfinished work in progress.
Change-Id: I978755d73630b8653b169a53f937c1332799e22e
|
|
Change-Id: Icc281ece80b81cdd95b6660c6a82077d283f5a62
|
|
Change-Id: I6f8c6827c00db50184a46f39968f882b944d18d4
Reviewed-on: https://gerrit.libreoffice.org/6967
Reviewed-by: Michael Stahl <mstahl@redhat.com>
Tested-by: LibreOffice gerrit bot <gerrit@libreoffice.org>
|
|
scopencl is only built when ENABLE_OPENCL is true, but the unit test
should still run successfully using the fallback mechanism without
scopencl (i.e. when --disable-opencl is used).
Change-Id: I4b44148f1f59ad8b3d9c78c2fd0e1cbe2030db37
|
|
Change-Id: I34c512573fcee108ab9b27a631671587665aad22
|
|
Change-Id: I4d98e88c4fcdf8c6e5ecbf31bc181dd186a85c9b
|
|
Test the jitting compiler with a simple vector formula like C1=A1+B1*2
This stresses:
1) vector operations add, mul and,
2) simple constant and vector operands.
|