Summary: | dev-dotnet/libgdiplus-2.10.9 - Variable 'libjpeg_prefix' not defined in '/usr/lib64/pkgconfig/libgdiplus.pc' | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | Bartosz Brachaczek <b.brachaczek> |
Component: | Current packages | Assignee: | dotnet project <dotnet> |
Status: | RESOLVED FIXED | ||
Severity: | normal | CC: | cynede, freedesktop-bugs, ianh, ostroffjh, rich0, rossi.f |
Priority: | Normal | Keywords: | NeedPatch |
Version: | unspecified | ||
Hardware: | All | ||
OS: | Linux | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- | |
Bug Depends on: | |||
Bug Blocks: | 487442 | ||
Attachments: | libgdiplus-2.10.9-configure.patch |
Description
Bartosz Brachaczek
2012-08-19 00:39:22 UTC
Will ask pkgconfig maintainers to try to know why .pc file is being wrongly generated (In reply to comment #1) > Will ask pkgconfig maintainers to try to know why .pc file is being wrongly > generated It's just a bug in their configure.in. They export -L${libjpeg_prefix}/lib in GDIPLUS_PKG_LIBS for use by libgdiplus.pc.in, and they don't define libjpeg_prefix in the .pc file. And they have very similar code for handling tiff and libgif, yet they don't include -L${libtiff_prefix}/lib nor -L${libgif_prefix}/lib in their .pc file. All their autoconf code is broken, it's need a substantial rewrite. Even the .pc files are broken, there is no excuse for libgdiplus.pc to export -ljpeg -lgif etc. in Libs. For Gentoo simply sed'ing the generated libgdiplus.pc to remove '-L${libjpeg_prefix}/lib' will be fine. I wonder if the priority or importance of this needs to be increased, as it has a secondary effect. Running "pkg-config --list-all" aborts on generating that error - so it outputs an abbreviated list of packages. It's certainly not critical, but I did waste some time wondering why "pkg-config --list-all" produced a much shorter list than the number of .pc files in the pkgconfig directory. (Yes, manually applying the fix from Comment #2 did fix the problem for me.) (In reply to comment #3) > I wonder if the priority or importance of this needs to be increased, as it > has a secondary effect. Running "pkg-config --list-all" aborts on > generating that error - so it outputs an abbreviated list of packages. It's > certainly not critical, but I did waste some time wondering why "pkg-config > --list-all" produced a much shorter list than the number of .pc files in the > pkgconfig directory. (Yes, manually applying the fix from Comment #2 did > fix the problem for me.) with what, dev-util/pkgconfig, dev-util/pkgconf, or something else? Samuli - thanks for making me think. :-) I suppose I should reword my question as asking whether there is any point in opening a separate bug against dev-util/pkgconfig so a problem in a .pc file (such as described in this bug) does not cause pkg-config to fail. (I don't use pkgconf, so I don't know whether it has the same problem or not.) Would be better to try to patch libgdiplus.pc.in than sedding .pc file after it being wrongly generated during building Created attachment 365050 [details, diff]
libgdiplus-2.10.9-configure.patch
here is a simple patch, the ebuild needs of course a run of eautoconf
Still not fixed in the -r1 ebuild, which is stable, nor the -r2, which is still ~. I filed bug 533146 against pkgconfig, to see if it can be made to simplly ignore invalid pc files. Its two years since someone provided a patch and there has not been any action by the .NET maintainers. Is this package orphaned or is the Gentoo .NET project orphaned? with a new gdiplus becoming stable soon I hope this problem just 'goes away' https://bugs.gentoo.org/show_bug.cgi?id=570434 that version is stable now |