Bug#806324: qgis: FTBFS on testing (unrecognized relocation in linking stage)
On 26-11-15 21:03, Sebastiaan Couwenberg wrote: > qgis indeed fails to build in testing, and testing only. > > I have no clue what's causing it, but it's I'm pretty sure it's not qgis. Thanks to Aurelien Jarno in #807170 the cause is now clear, to quote: " The problem is just that libQtUiTools.a has been built with the binutils from unstable, which introduce new relocation types. The .o object files (and thus static libraries) generated by binutils from unstable therefore can't be processed by the one from testing. " https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=807170#19 If everything goes well, binutils (2.25.90.20151209-1) will migrate to testing in 4 days. Let's hope nothing prevents that. Kind Regards, Bas -- GPG Key ID: 4096R/6750F10AE88D4AF1 Fingerprint: 8182 DE41 7056 408D 6146 50D1 6750 F10A E88D 4AF1
Bug#806324: qgis: FTBFS on testing (unrecognized relocation in linking stage)
Hello, I confirm Yorik's analysis: I have the same issue while compiling Freecad from mainstream sources on my testing. I downgraded the following packages to these versions: libqt4-dbus_4.8.7+dfsg-3_amd64.deb libqt4-declarative_4.8.7+dfsg-3_amd64.deb libqt4-designer_4.8.7+dfsg-3_amd64.deb libqt4-dev_4.8.7+dfsg-3_amd64.deb libqt4-dev-bin_4.8.7+dfsg-3_amd64.deb libqt4-help_4.8.7+dfsg-3_amd64.deb libqt4-network_4.8.7+dfsg-3_amd64.deb libqt4-qt3support_4.8.7+dfsg-3_amd64.deb libqt4-script_4.8.7+dfsg-3_amd64.deb libqt4-scripttools_4.8.7+dfsg-3_amd64.deb libqt4-sql_4.8.7+dfsg-3_amd64.deb libqt4-svg_4.8.7+dfsg-3_amd64.deb libqt4-test_4.8.7+dfsg-3_amd64.deb libqt4-xml_4.8.7+dfsg-3_amd64.deb libqt4-xmlpatterns_4.8.7+dfsg-3_amd64.deb libqtcore4_4.8.7+dfsg-3_amd64.deb libqtdbus4_4.8.7+dfsg-3_amd64.deb libqtgui4_4.8.7+dfsg-3_amd64.deb qdbus_4.8.7+dfsg-3_amd64.deb qt4-linguist-tools_4.8.7+dfsg-3_amd64.deb qt4-qmake_4.8.7+dfsg-3_amd64.deb And after that the build worked. Regard, Olivier
Bug#806324: qgis: FTBFS on testing (unrecognized relocation in linking stage)
On 29-11-15 16:20, Yorik van Havre wrote: > Apparently something changed or is buggy in latest libqt4-dev (which contains > libQtUiTools.a). That's my impression too. The problem is not qgis, it's in its dependencies. > Are you sure that qgis still builds with unstable? Yes, as I wrote in message #10: " Have you verified that it affects unstable too? Because my unstable builds succeed just fine. " https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=806324#10 > It seems to me that all the Qt4 stuff is now at same version > between testing and unstable... > > Probably this bug is actually in one of the Qt4 packages... Or the interaction with its dependencies, qt4 should be the same in testing and unstable, its dependencies in turn may not be. Kind Regards, Bas -- GPG Key ID: 4096R/6750F10AE88D4AF1 Fingerprint: 8182 DE41 7056 408D 6146 50D1 6750 F10A E88D 4AF1
Bug#806324: qgis: FTBFS on testing (unrecognized relocation in linking stage)
FreeCAD seems to suffer from the exact same problem since a couple of days: [ 15%] Linking CXX shared library ../../lib/libFreeCADGui.so /usr/bin/ld: /usr/lib/gcc/x86_64-linux-gnu/libQtUiTools.a(quiloader.o): unrecognized relocation (0x2a) in section `.text' /usr/bin/ld: final link failed: Bad value Apparently something changed or is buggy in latest libqt4-dev (which contains libQtUiTools.a). Are you sure that qgis still builds with unstable? It seems to me that all the Qt4 stuff is now at same version between testing and unstable... Probably this bug is actually in one of the Qt4 packages... Cheers Yorik
Bug#806324: qgis: FTBFS on testing (unrecognized relocation in linking stage)
On Thu, Nov 26, 2015 at 09:03:24PM +0100, Sebastiaan Couwenberg wrote: > Control: tags -1 - moreinfo unreproducible > Control: tags -1 + help > > Hi Santiago, > > qgis indeed fails to build in testing, and testing only. > > I have no clue what's causing it, but it's I'm pretty sure it's not qgis. > > Because I cannot fix a package in testing that is not broken in > unstable, I'm tempted to close this wontfix I would recommend that you don't. "RC" and "wontfix" do not mix very well. > but I'll ask for help first. That's a good idea indeed. > Until this issue also affects unstable, the serious severity is not > appropriate. It is. Remember that it's a snapshot of testing what we release as stable when a new stable is released, not a snapshot of unstable (so it would be quite the reverse of what you said: until an issue also affects testing, it is not RC). Don't worry, I will not start a severity war for this issue, but if you do not believe me, please ask any release manager. Thanks.
Bug#806324: qgis: FTBFS on testing (unrecognized relocation in linking stage)
FWIW, the diff between build dependencies is attached. qgis_build-deps.diff Description: inode/symlink
Bug#806324: qgis: FTBFS on testing (unrecognized relocation in linking stage)
Control: tags -1 - moreinfo unreproducible Control: tags -1 + help Hi Santiago, qgis indeed fails to build in testing, and testing only. I have no clue what's causing it, but it's I'm pretty sure it's not qgis. Because I cannot fix a package in testing that is not broken in unstable, I'm tempted to close this wontfix but I'll ask for help first. Until this issue also affects unstable, the serious severity is not appropriate. Kind Regards, Bas -- GPG Key ID: 4096R/6750F10AE88D4AF1 Fingerprint: 8182 DE41 7056 408D 6146 50D1 6750 F10A E88D 4AF1
Bug#806324: qgis: FTBFS on testing (unrecognized relocation in linking stage)
Package: src:qgis Version: 2.8.3+dfsg-5 User: reproducible-bui...@lists.alioth.debian.org Usertags: ftbfs Severity: serious Dear maintainer: This package fails to build from source in testing/amd64. Follows a snippet from the build log: [...] cd /<>/qgis-2.8.3+dfsg/obj-x86_64-linux-gnu/src/gui && /usr/bin/c++ -DQT_CORE_LIB -DQT_GUI_LIB -DQT_NETWORK_LIB -DQT_NO_CAST_TO_ASCII -DQT_NO_DEBUG -DQT_OPENGL_LIB -DQT_SQL_LIB -DQT_SVG_LIB -DQT_WEBKIT_LIB -DQT_XML_LIB -Dqgis_gui_EXPORTS -g -O2 -fstack-protector-strong -Wformat -Werror=format-security -D_FORTIFY_SOURCE=2 -O2 -D_FORTIFY_SOURCE=2 -DSPATIALITE_VERSION_GE_4_0_0 -DSPATIALITE_VERSION_G_4_1_1 -DSPATIALITE_HAS_INIT_EX -std=c++11 -Wall -Wextra -Wno-long-long -Wformat-security -Wno-strict-aliasing -fPIC -isystem /usr/include/qt4 -isystem /usr/include/qt4/QtOpenGL -isystem /usr/include/qt4/QtSvg -isystem /usr/include/qt4/QtWebKit -isystem /usr/include/qt4/QtGui -isystem /usr/include/qt4/QtXml -isystem /usr/include/qt4/QtSql -isystem /usr/include/qt4/QtNetwork -isystem /usr/include/qt4/QtCore -I/<>/qgis-2.8.3+dfsg/obj-x86_64-linux-gnu -I/usr/include/qt4/QtUiTools -I/<>/qgis-2.8.3+dfsg/src/gui -I/<>/qgis-2.8.3+dfsg/src/gui/symbology-ng - I/<>/qgis-2.8.3+dfsg/src/gui/attributetable -I/<>/qgis-2.8.3+dfsg/src/gui/editorwidgets -I/<>/qgis-2.8.3+dfsg/src/gui/editorwidgets/core -I/<>/qgis-2.8.3+dfsg/src/gui/layertree -I/<>/qgis-2.8.3+dfsg/src/gui/../core -I/<>/qgis-2.8.3+dfsg/src/gui/../core/composer -I/<>/qgis-2.8.3+dfsg/src/gui/../core/layertree -I/<>/qgis-2.8.3+dfsg/src/gui/../core/raster -I/<>/qgis-2.8.3+dfsg/src/gui/../core/symbology-ng -I/<>/qgis-2.8.3+dfsg/obj-x86_64-linux-gnu/src/gui -I/<>/qgis-2.8.3+dfsg/obj-x86_64-linux-gnu/src/gui/../ui -I/usr/include/qwt -I/usr/include/gdal-DCORE_EXPORT= -DGUI_EXPORT= -DPYTHON_EXPORT= -DANALYSIS_EXPORT= -DAPP_EXPORT= -DCUSTOMWIDGETS_EXPORT= -DSERVER_EXPORT= -Wno-deprecated-declarations -o CMakeFiles/qgis_gui.dir/layertree/moc_qgslayertreeviewdefaultactions.cxx.o -c /<>/qgis-2.8.3+dfsg/obj-x86_64-linux-gnu/src/gui/layertree/moc_qgslayertreeviewdefaultactio ns.cxx [ 43%] Linking CXX shared library ../../output/lib/libqgis_gui.so cd /<>/qgis-2.8.3+dfsg/obj-x86_64-linux-gnu/src/gui && /usr/bin/cmake -E cmake_link_script CMakeFiles/qgis_gui.dir/link.txt --verbose=1 /usr/bin/c++ -fPIC -g -O2 -fstack-protector-strong -Wformat -Werror=format-security -D_FORTIFY_SOURCE=2 -O2 -D_FORTIFY_SOURCE=2 -DSPATIALITE_VERSION_GE_4_0_0 -DSPATIALITE_VERSION_G_4_1_1 -DSPATIALITE_HAS_INIT_EX -std=c++11 -Wall -Wextra -Wno-long-long -Wformat-security -Wno-strict-aliasing -Wl,-z,relro -Wl,--no-undefined -shared -Wl,-soname,libqgis_gui.so.2.8.3 -o ../../output/lib/libqgis_gui.so.2.8.3 CMakeFiles/qgis_gui.dir/raster/qgsrasterminmaxwidget.cpp.o CMakeFiles/qgis_gui.dir/raster/qgsrasterrendererwidget.cpp.o CMakeFiles/qgis_gui.dir/raster/qgsmultibandcolorrendererwidget.cpp.o CMakeFiles/qgis_gui.dir/raster/qgspalettedrendererwidget.cpp.o CMakeFiles/qgis_gui.dir/raster/qgssinglebandgrayrendererwidget.cpp.o CMakeFiles/qgis_gui.dir/raster/qgssinglebandpseudocolorrendererwidget.cpp.o CMakeFiles/qgis_gui.dir/raster/qgsrasterhistogramwidget.cpp.o CMakeFiles/qgis_gui.dir/symbology-ng/qgsbrushstylecombobox.cpp.o CMakeFiles/qgis_gui.dir/symbology-ng/qgscolorrampcombobox.cpp.o CM akeFiles/qgis_gui.dir/symbology-ng/qgsdashspacedialog.cpp.o CMakeFiles/qgis_gui.dir/symbology-ng/qgsdatadefinedsymboldialog.cpp.o CMakeFiles/qgis_gui.dir/symbology-ng/qgspenstylecombobox.cpp.o CMakeFiles/qgis_gui.dir/symbology-ng/qgssymbollayerv2widget.cpp.o CMakeFiles/qgis_gui.dir/symbology-ng/qgsrendererv2widget.cpp.o CMakeFiles/qgis_gui.dir/symbology-ng/qgssinglesymbolrendererv2widget.cpp.o CMakeFiles/qgis_gui.dir/symbology-ng/qgscategorizedsymbolrendererv2widget.cpp.o CMakeFiles/qgis_gui.dir/symbology-ng/qgsgraduatedsymbolrendererv2widget.cpp.o CMakeFiles/qgis_gui.dir/symbology-ng/qgsrulebasedrendererv2widget.cpp.o CMakeFiles/qgis_gui.dir/symbology-ng/qgsheatmaprendererwidget.cpp.o CMakeFiles/qgis_gui.dir/symbology-ng/qgsinvertedpolygonrendererwidget.cpp.o CMakeFiles/qgis_gui.dir/symbology-ng/qgsrendererv2propertiesdialog.cpp.o CMakeFiles/qgis_gui.dir/symbology-ng/qgsstylev2managerdialog.cpp.o CMakeFiles/qgis_gui.dir/symbology-ng/qgssymbollevelsv2dialog.cpp.o CMakeFiles/qgis_gui .dir/symbology-ng/qgssymbolv2selectordialog.cpp.o CMakeFiles/qgis_gui.dir/symbology-ng/qgsvectorgradientcolorrampv2dialog.cpp.o CMakeFiles/qgis_gui.dir/symbology-ng/qgsvectorrandomcolorrampv2dialog.cpp.o CMakeFiles/qgis_gui.dir/symbology-ng/qgsvectorcolorbrewercolorrampv2dialog.cpp.o CMakeFiles/qgis_gui.dir/symbology-ng/qgscptcitycolorrampv2dialog.cpp.o CMakeFiles/qgis_gui.dir/symbology-ng/characterwidget.cpp.o CMakeFiles/qgis_gui.dir/symbology-ng/qgsstylev2exportimportdialog.cpp.o
Bug#806324: qgis: FTBFS on testing (unrecognized relocation in linking stage)
Control: severity -1 important Control: tags -1 unreproducible moreinfo Hi Santiago, Thanks for your work on reproducible builds. On 2015-11-26 16:06, Santiago Vila wrote: This package fails to build from source in testing/amd64. Why are your reporting build failures for testing? Have you verified that it affects unstable too? Because my unstable builds succeed just fine. Kind Regards, Bas
Bug#806324: qgis: FTBFS on testing (unrecognized relocation in linking stage)
On Thu, Nov 26, 2015 at 05:14:26PM +0100, Bas Couwenberg wrote: > Control: severity -1 important > Control: tags -1 unreproducible moreinfo > > Hi Santiago, > > Thanks for your work on reproducible builds. > > On 2015-11-26 16:06, Santiago Vila wrote: > >This package fails to build from source in testing/amd64. > > Why are your reporting build failures for testing? Well, because packages in testing should be buildable from testing. Otherwise, when we release testing as the new stable, we would have a package in stable which fails to build in stable. [ Please note that I would not have filed this bug if testing and unstable had different versions, because in such case the only interesting version to check would be the version in unstable. In this case we are talking about version 2.8.3+dfsg-5 which is already in testing and for which there is not a newer version in unstable to check. ] > Have you verified that it affects unstable too? Because my unstable > builds succeed just fine. It probably builds in unstable, yes, but what I'm reporting here is not that it fails to build in a "generic sense". I'm reporting that it fails to build in testing. So, before tagging this as "unreproducible", could you please try to build it on testing, which is what this report is about? Thanks.
Bug#806324: qgis: FTBFS on testing (unrecognized relocation in linking stage)
On 2015-11-26 18:10, Santiago Vila wrote: On Thu, Nov 26, 2015 at 05:14:26PM +0100, Bas Couwenberg wrote: Have you verified that it affects unstable too? Because my unstable builds succeed just fine. It probably builds in unstable, yes, but what I'm reporting here is not that it fails to build in a "generic sense". I'm reporting that it fails to build in testing. So, before tagging this as "unreproducible", could you please try to build it on testing, which is what this report is about? I don't really see how that's useful, packages in testing don't get rebuilt. Or do you suggest to uploaded to testing-proposed-uploads with a fix for this testing-only issue? Kind Regards, Bas
Bug#806324: qgis: FTBFS on testing (unrecognized relocation in linking stage)
On Thu, Nov 26, 2015 at 06:16:30PM +0100, Bas Couwenberg wrote: > On 2015-11-26 18:10, Santiago Vila wrote: > >On Thu, Nov 26, 2015 at 05:14:26PM +0100, Bas Couwenberg wrote: > >>Have you verified that it affects unstable too? Because my unstable > >>builds succeed just fine. > > > >It probably builds in unstable, yes, but what I'm reporting here is > >not that it fails to build in a "generic sense". I'm reporting that it > >fails to build in testing. > > > >So, before tagging this as "unreproducible", could you please try to > >build it on testing, which is what this report is about? > > I don't really see how that's useful, Well, there is GPL compliance, the Social Contract, you know. It would be quite useless that we provide a free operating system complete with source code if users are not able to build packages from the source we provide. At the very minimum, building it on testing would allow you to check if the unreproducible tag does really apply here or not. (I think it does not apply because I have two different build logs which fail on testing). > packages in testing don't get rebuilt. This is not completely true. In some cases release managers perform binary-only NMUs. The typical case are transitions, where packages are linked to new versions of libraries without any real source code change. But I don't think that's really relevant in this case. The issue here is a basic principle in Debian: we want everybody to be able to rebuild packages, regardless of the fact that we do it as well or not. > Or do you suggest to uploaded to testing-proposed-uploads > with a fix for this testing-only issue? No. If you find a way to modify the source so that the new package builds ok in both testing and unstable, you would just upload it to unstable. The fixed package would propagate to testing and then everything would work in both testing and unstable. What I suggest is that you investigate about the reasons why this problem happens. Restoring the initial severity would also be nice. This is a FTBFS bug and the consensus is that those bugs are of "serious" severity. As for the tags, I don't see what "moreinfo" you need. I said "testing" right from the beginning. Thanks.