Re: Trying to solve a crasher in editeng unit test
On Thu, 2012-04-12 at 10:14 -0400, Kohei Yoshida wrote: On Thu, Apr 12, 2012 at 10:06 AM, Caolán McNamara caol...@redhat.com wrote: *mumable* unotest::testDown has its dispose commented out ages ago, putting that back in shows some other problems with various mixes of global singletons that exist for the duration of a test-suite and the per-test globals which get setup and reset for each test. Yes. Some of my changes were an attempt to address that ok, plan b, reorganize matters so that we bootstrap the initial uno component context once before all tests, and dispose it once after all tests are complete. Rather than setup/teardown before/after each test which just doesn't fly so well. reenabled editeng test again, and await inevitable fallout on some platform or other. C. ___ LibreOffice mailing list LibreOffice@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice
Re: Trying to solve a crasher in editeng unit test
On Fri, Apr 13, 2012 at 8:15 AM, Caolán McNamara caol...@redhat.com wrote: Yes. Some of my changes were an attempt to address that ok, plan b, reorganize matters so that we bootstrap the initial uno component context once before all tests, and dispose it once after all tests are complete. Rather than setup/teardown before/after each test which just doesn't fly so well. Thanks Caolan, for solving this. I just didn't realize how tangled up the life cycle of these singleton objects was... reenabled editeng test again, and await inevitable fallout on some platform or other. Keeping my fingers crossed. :-) Kohei ___ LibreOffice mailing list LibreOffice@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice
Re: Trying to solve a crasher in editeng unit test
On Fri, 2012-04-13 at 12:14 -0400, Kohei Yoshida wrote: On Fri, Apr 13, 2012 at 8:15 AM, Caolán McNamara caol...@redhat.com wrote: reenabled editeng test again, and await inevitable fallout on some platform or other. Keeping my fingers crossed. :-) Inevitable platform fallout on windows happened, now fixed I think. At least the windows tinderboxs stopped bleating at me. C. ___ LibreOffice mailing list LibreOffice@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice
Re: Trying to solve a crasher in editeng unit test
On 04/11/2012 10:44 PM, Kohei Yoshida wrote: On Wed, 2012-04-11 at 21:29 +0100, Caolán McNamara wrote: On Wed, 2012-04-11 at 15:56 -0400, Kohei Yoshida wrote: Hey there, I'm trying to set up a new unit test for the edit engine core, but struggling with one crasher. http://cgit.freedesktop.org/libreoffice/core/commit/?id=6089be1a7a17ae849d8793520e9d8d36296ed5cf should get you closer anyway, and into a more mundane crash with SfxItemPools Thanks a lot. This helps. Hopefully any fix of yours will also fix the following problem: My local --enable-dbgutil make check build keeps failing with a crash in sw_subsequent_rtftok. Valgrind gives ==32152== Invalid read of size 8 ==32152==at 0x12DC7689: SfxItemPool::GetDefaultItem(unsigned short) const (/svl/source/items/itempool.cxx:874) ==32152==by 0x12DC6058: SfxItemPool::Put(SfxPoolItem const, unsigned short) (/svl/source/items/itempool.cxx:681) ==32152==by 0x12DC5E44: SfxItemPool::Put(SfxPoolItem const, unsigned short) (/svl/source/items/itempool.cxx:663) ==32152==by 0x12DDD519: SfxItemSet::Put(SfxPoolItem const, unsigned short) (/svl/source/items/itemset.cxx:609) ==32152==by 0x213CB7E5: SfxItemSet::Put(SfxPoolItem const) (/solver/unxlngx6/inc/svl/itemset.hxx:150) ==32152==by 0x214DDDA4: sdr::properties::TextProperties::ForceDefaultAttributes() (/svx/source/sdr/properties/textproperties.cxx:384) ==32152==by 0x214DBC82: sdr::properties::DefaultProperties::GetObjectItemSet() const (/svx/source/sdr/properties/defaultproperties.cxx:94) ==32152==by 0x215A02D1: SdrObject::GetObjectItemSet() const (/svx/source/svdraw/svdobj.cxx:1981) ==32152==by 0x215FB757: SdrTextObj::GetTextLeftDistance() const (/svx/source/svdraw/svdotext.cxx:1792) ==32152==by 0x2161D645: SdrTextObj::NbcResize(Point const, Fraction const, Fraction const) (/svx/source/svdraw/svdotxtr.cxx:125) ==32152==by 0x215ED66E: SdrPathObj::NbcResize(Point const, Fraction const, Fraction const) (/svx/source/svdraw/svdopath.cxx:2370) ==32152==by 0x215EDF95: SdrPathObj::NbcSetSnapRect(Rectangle const) (/svx/source/svdraw/svdopath.cxx:2466) ==32152==by 0x2159F3A1: SdrObject::SetSnapRect(Rectangle const) (/svx/source/svdraw/svdobj.cxx:1687) ==32152==by 0x217086D4: SvxDrawPage::_CreateSdrObject(com::sun::star::uno::Referencecom::sun::star::drawing::XShape const) (/svx/source/unodraw/unopage.cxx:518) ==32152==by 0x2185BF74: SvxFmDrawPage::_CreateSdrObject(com::sun::star::uno::Referencecom::sun::star::drawing::XShape const) (/svx/source/form/fmdpage.cxx:102) ==32152==by 0x1F39D352: SwFmDrawPage::_CreateSdrObject(com::sun::star::uno::Referencecom::sun::star::drawing::XShape const) (/sw/source/core/unocore/unodraw.cxx:333) ==32152==by 0x21709CEF: SvxDrawPage::CreateSdrObject(com::sun::star::uno::Referencecom::sun::star::drawing::XShape const) (/svx/source/unodraw/unopage.cxx:850) ==32152==by 0x217071B0: SvxDrawPage::add(com::sun::star::uno::Referencecom::sun::star::drawing::XShape const) (/svx/source/unodraw/unopage.cxx:246) ==32152==by 0x1F39ED4F: SwXDrawPage::add(com::sun::star::uno::Referencecom::sun::star::drawing::XShape const) (/sw/source/core/unocore/unodraw.cxx:630) ==32152==by 0x255DA9D7: writerfilter::rtftok::RTFDocumentImpl::popState() (/writerfilter/source/rtftok/rtfdocumentimpl.cxx:3299) ==32152==by 0x25614308: writerfilter::rtftok::RTFTokenizer::resolveParse() (/writerfilter/source/rtftok/rtftokenizer.cxx:110) ==32152==by 0x255C4960: writerfilter::rtftok::RTFDocumentImpl::resolve(writerfilter::Stream) (/writerfilter/source/rtftok/rtfdocumentimpl.cxx:548) ==32152==by 0x231E0722: RtfFilter::filter(com::sun::star::uno::Sequencecom::sun::star::beans::PropertyValue const) (/writerfilter/source/filter/RtfFilter.cxx:99) ==32152==by 0x1C359018: SfxObjectShell::ImportFrom(SfxMedium, bool) (/sfx2/source/doc/objstor.cxx:2238) ==32152==by 0x1C351224: SfxObjectShell::DoLoad(SfxMedium*) (/sfx2/source/doc/objstor.cxx:730) ==32152==by 0x1C39E867: SfxBaseModel::load(com::sun::star::uno::Sequencecom::sun::star::beans::PropertyValue const) (/sfx2/source/doc/sfxbasemodel.cxx:1900) ==32152==by 0x1C3EC76C: SfxFrameLoader_Impl::load(com::sun::star::uno::Sequencecom::sun::star::beans::PropertyValue const, com::sun::star::uno::Referencecom::sun::star::frame::XFrame const) (/sfx2/source/view/frmload.cxx:611) ==32152==by 0x1A9C5EC5: framework::LoadEnv::impl_loadContent() (/framework/source/loadenv/loadenv.cxx:1160) ==32152==by 0x1A9C2696: framework::LoadEnv::startLoading() (/framework/source/loadenv/loadenv.cxx:418) ==32152==by 0x1A9C150C: framework::LoadEnv::loadComponentFromURL(com::sun::star::uno::Referencecom::sun::star::frame::XComponentLoader const, com::sun::star::uno::Referencecom::sun::star::lang::XMultiServiceFactory const, rtl::OUString const, rtl::OUString const, int, com::sun::star::uno::Sequencecom::sun::star::beans::PropertyValue const)
Re: Trying to solve a crasher in editeng unit test
On Thu, 2012-04-12 at 10:50 +0200, Stephan Bergmann wrote: On 04/11/2012 10:44 PM, Kohei Yoshida wrote: On Wed, 2012-04-11 at 21:29 +0100, Caolán McNamara wrote: On Wed, 2012-04-11 at 15:56 -0400, Kohei Yoshida wrote: Hey there, I'm trying to set up a new unit test for the edit engine core, but struggling with one crasher. http://cgit.freedesktop.org/libreoffice/core/commit/?id=6089be1a7a17ae849d8793520e9d8d36296ed5cf should get you closer anyway, and into a more mundane crash with SfxItemPools Thanks a lot. This helps. Hopefully any fix of yours will also fix the following problem: My local --enable-dbgutil make check build keeps failing with a crash in sw_subsequent_rtftok. Valgrind gives Crap. Maybe a unit test in editeng wasn't such a good idea after all. Kohei ___ LibreOffice mailing list LibreOffice@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice
Re: Trying to solve a crasher in editeng unit test
On Thu, 2012-04-12 at 09:46 -0400, Kohei Yoshida wrote: Crap. Maybe a unit test in editeng wasn't such a good idea after all. I disabled it and backed out a few bit for the moment. I have a fairly good idea what the various problems are so I might have a go at fixing it up again in a bit. *mumable* unotest::testDown has its dispose commented out ages ago, putting that back in shows some other problems with various mixes of global singletons that exist for the duration of a test-suite and the per-test globals which get setup and reset for each test. C. ___ LibreOffice mailing list LibreOffice@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice
Re: Trying to solve a crasher in editeng unit test
On Thu, Apr 12, 2012 at 10:06 AM, Caolán McNamara caol...@redhat.com wrote: On Thu, 2012-04-12 at 09:46 -0400, Kohei Yoshida wrote: Crap. Maybe a unit test in editeng wasn't such a good idea after all. I disabled it and backed out a few bit for the moment. I have a fairly good idea what the various problems are so I might have a go at fixing it up again in a bit. *mumable* unotest::testDown has its dispose commented out ages ago, putting that back in shows some other problems with various mixes of global singletons that exist for the duration of a test-suite and the per-test globals which get setup and reset for each test. Yes. Some of my changes were an attempt to address that. But apparently my skill is not up for it. I give up now. You did the right thing by backing out my changes. Kohei ___ LibreOffice mailing list LibreOffice@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice
Re: Trying to solve a crasher in editeng unit test
On Wed, 2012-04-11 at 15:56 -0400, Kohei Yoshida wrote: Hey there, I'm trying to set up a new unit test for the edit engine core, but struggling with one crasher. http://cgit.freedesktop.org/libreoffice/core/commit/?id=6089be1a7a17ae849d8793520e9d8d36296ed5cf should get you closer anyway, and into a more mundane crash with SfxItemPools C. ___ LibreOffice mailing list LibreOffice@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libreoffice