Re: WIP: print/scribus (segmentation fault)
On Mon, Feb 03, 2020 at 09:14:19AM +0100, Landry Breuil wrote: > On Mon, Feb 03, 2020 at 07:05:38AM +0100, Rafael Sadowski wrote: > > On Sun Feb 02, 2020 at 07:29:56PM -0800, Nam Nguyen wrote: > > > >>> > Rafael Sadowski writes: > > > > > > To test the port, I followed the tutorial in Help > Scribus Manual > > > > Scribus Basics > Quick Start Guide. I was able to make this by following > > > the tutorial: https://namtsui.com/public/rembrandt.jpg > > > > > > Feedback and tests are welcome. > > > > > > > Works like a charm. All plugins are found. Tested with the PDF plugin. > > I'm not a scribus user but it feels stable. Thanks Nam Nguyen, great > > work! > > I concur, many thanks nam for digging into this and coming with fixes ! > At the time i tried updating it it, didnt get that far.. > > Very happy to see this port finally updated :) The updated has finally been commited, thanks to all involved :) Landry
Re: WIP: print/scribus (segmentation fault)
On Mon, Feb 03, 2020 at 07:05:38AM +0100, Rafael Sadowski wrote: > On Sun Feb 02, 2020 at 07:29:56PM -0800, Nam Nguyen wrote: > > >>> > Rafael Sadowski writes: > > >>> > > > >>> > > Here is an update for scribus to the newest qt5 version. Long time > > >>> > > ago > > >>> > > Landry tried to update scribus without response from ports@. > > >>> > > > > >>> > > Here is a new try. Unfortunately this version crashes immediately > > >>> > > without a backtrace. > > >>> > > > > >>> > > Anyone want to take a look at this? I would be very grateful. > > >>> > > > > >>> > > Diff below includes patches from ArchLinux to unbreak the build with > > >>> > > our popper. > > > > I managed to get scribus to work. Here is a diff that additionally: > > - deletes DOS line endings using Stuart's hint > > - fixes segfault with patches/patch-scribus_ui_colorlistmodel_{cpp,h} > > > > First, I enabled debugging in the port. > > DEBUG = -g > > CFLAGS =-g > > CXXFLAGS = -g > > > > Then, I installed debug-qtbase. The debug packages were helpful. > > $ pkg_add debug-qtbase > > > > The debugger shows: > > (gdb) run > > Starting program: /usr/local/bin/scribus > > > > Program received signal SIGSEGV, Segmentation fault. > > 0x099ac44e44e2 in std::__1::__atomic_base > false>::load(std::__1::memory_order) const () from > > /usr/local/lib/libQt5Core.so.2.2 > > (gdb) bt > > #0 0x099ac44e44e2 in std::__1::__atomic_base > false>::load(std::__1::memory_order) const () from > > /usr/local/lib/libQt5Core.so.2.2 > > #1 0x099ac44e4471 in int > > QAtomicOps::load(std::__1::atomic const&) () from > > /usr/local/lib/libQt5Core.so.2.2 > > #2 0x099ac44e442f in QBasicAtomicInteger::load() const () > >from /usr/local/lib/libQt5Core.so.2.2 > > #3 0x099ac44e5176 in QtPrivate::RefCount::ref() () > >from /usr/local/lib/libQt5Core.so.2.2 > > #4 0x099ac462fde1 in QString::operator=(QString const&) () > >from /usr/local/lib/libQt5Core.so.2.2 > > #5 0x0997f478be60 in _GLOBAL__sub_I_colorlistmodel.cpp () > > at > > /usr/ports/pobj/scribus-1.5.5/scribus-1.5.5/scribus/ui/colorlistmodel.cpp:13 > > > > colorlistmodel.cpp:13 reads: > > ColorPixmapValue ColorListModel::m_NoneColor(ScColor(), nullptr, > > CommonStrings::None); > > > > in colorlistmodel.h: > > //! Dummy object used to represent None Color > > static ColorPixmapValue m_NoneColor; > > > > Given frame 4's QString, I think the real problem is with the usage of > > CommonStrings::None. I noticed that m_NoneColor was not actually used in > > ui/colorlistmodel.{h,cpp} or anywhere else in the project. I just > > deleted m_NoneColor and that actually resolved the segfault. > > > > I can poke at it some more to see if there is a more optimal solution, > > but at least the program starts up now. If this is sufficient, it should > > be ready for testing. > > > > To test the port, I followed the tutorial in Help > Scribus Manual > > > Scribus Basics > Quick Start Guide. I was able to make this by following > > the tutorial: https://namtsui.com/public/rembrandt.jpg > > > > Feedback and tests are welcome. > > > > Works like a charm. All plugins are found. Tested with the PDF plugin. > I'm not a scribus user but it feels stable. Thanks Nam Nguyen, great > work! I concur, many thanks nam for digging into this and coming with fixes ! At the time i tried updating it it, didnt get that far.. Very happy to see this port finally updated :) Landry
Re: WIP: print/scribus (segmentation fault)
On Sun Feb 02, 2020 at 07:29:56PM -0800, Nam Nguyen wrote: > >>> > Rafael Sadowski writes: > >>> > > >>> > > Here is an update for scribus to the newest qt5 version. Long time ago > >>> > > Landry tried to update scribus without response from ports@. > >>> > > > >>> > > Here is a new try. Unfortunately this version crashes immediately > >>> > > without a backtrace. > >>> > > > >>> > > Anyone want to take a look at this? I would be very grateful. > >>> > > > >>> > > Diff below includes patches from ArchLinux to unbreak the build with > >>> > > our popper. > > I managed to get scribus to work. Here is a diff that additionally: > - deletes DOS line endings using Stuart's hint > - fixes segfault with patches/patch-scribus_ui_colorlistmodel_{cpp,h} > > First, I enabled debugging in the port. > DEBUG = -g > CFLAGS =-g > CXXFLAGS = -g > > Then, I installed debug-qtbase. The debug packages were helpful. > $ pkg_add debug-qtbase > > The debugger shows: > (gdb) run > Starting program: /usr/local/bin/scribus > > Program received signal SIGSEGV, Segmentation fault. > 0x099ac44e44e2 in std::__1::__atomic_base false>::load(std::__1::memory_order) const () from > /usr/local/lib/libQt5Core.so.2.2 > (gdb) bt > #0 0x099ac44e44e2 in std::__1::__atomic_base false>::load(std::__1::memory_order) const () from > /usr/local/lib/libQt5Core.so.2.2 > #1 0x099ac44e4471 in int > QAtomicOps::load(std::__1::atomic const&) () from > /usr/local/lib/libQt5Core.so.2.2 > #2 0x099ac44e442f in QBasicAtomicInteger::load() const () >from /usr/local/lib/libQt5Core.so.2.2 > #3 0x099ac44e5176 in QtPrivate::RefCount::ref() () >from /usr/local/lib/libQt5Core.so.2.2 > #4 0x099ac462fde1 in QString::operator=(QString const&) () >from /usr/local/lib/libQt5Core.so.2.2 > #5 0x0997f478be60 in _GLOBAL__sub_I_colorlistmodel.cpp () > at > /usr/ports/pobj/scribus-1.5.5/scribus-1.5.5/scribus/ui/colorlistmodel.cpp:13 > > colorlistmodel.cpp:13 reads: > ColorPixmapValue ColorListModel::m_NoneColor(ScColor(), nullptr, > CommonStrings::None); > > in colorlistmodel.h: > //! Dummy object used to represent None Color > static ColorPixmapValue m_NoneColor; > > Given frame 4's QString, I think the real problem is with the usage of > CommonStrings::None. I noticed that m_NoneColor was not actually used in > ui/colorlistmodel.{h,cpp} or anywhere else in the project. I just > deleted m_NoneColor and that actually resolved the segfault. > > I can poke at it some more to see if there is a more optimal solution, > but at least the program starts up now. If this is sufficient, it should > be ready for testing. > > To test the port, I followed the tutorial in Help > Scribus Manual > > Scribus Basics > Quick Start Guide. I was able to make this by following > the tutorial: https://namtsui.com/public/rembrandt.jpg > > Feedback and tests are welcome. > Works like a charm. All plugins are found. Tested with the PDF plugin. I'm not a scribus user but it feels stable. Thanks Nam Nguyen, great work! RS
Re: WIP: print/scribus (segmentation fault)
On 2020/02/02 20:06, Rafael Sadowski wrote: > On Sun Feb 02, 2020 at 10:59:13AM -0800, Nam Nguyen wrote: > > Rafael Sadowski writes: > > > > > Here is an update for scribus to the newest qt5 version. Long time ago > > > Landry tried to update scribus without response from ports@. > > > > > > Here is a new try. Unfortunately this version crashes immediately > > > without a backtrace. > > > > > > Anyone want to take a look at this? I would be very grateful. > > > > > > Diff below includes patches from ArchLinux to unbreak the build with > > > our popper. > > > > Please reattach the diff because it has ^M endings. Please fix those up in the file in post-extract so the diff doesn't have to cope with these. post-extract: @cd ${WRKSRC} && perl -i -pe 's/\r$$//' $somefile
Re: WIP: print/scribus (segmentation fault)
Rafael Sadowski writes: > Here is an update for scribus to the newest qt5 version. Long time ago > Landry tried to update scribus without response from ports@. > > Here is a new try. Unfortunately this version crashes immediately > without a backtrace. > > Anyone want to take a look at this? I would be very grateful. > > Diff below includes patches from ArchLinux to unbreak the build with > our popper. Please reattach the diff because it has ^M endings. > > Best regards, > > Rafael