Summary: | app-office/libalkimia: stores incorrect path to libgmp or libmpir in LibAlkimia5Targets.cmake (was: app-office/kmymoney-5.0.4: error: '/usr/lib/libgmp.so', needed by 'lib/libkmm_mymoney.so.5.0.4', missing) | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | boxcars |
Component: | Current packages | Assignee: | Gentoo KDE team <kde> |
Status: | RESOLVED DUPLICATE | ||
Severity: | normal | CC: | Dan.Johansson, dflogeras2, jaak, joerg.schaible, jstein, ostroffjh |
Priority: | Normal | ||
Version: | unspecified | ||
Hardware: | AMD64 | ||
OS: | Linux | ||
Whiteboard: | solution: rebuild app-office/libalkimia | ||
Package list: | Runtime testing required: | --- | |
Bug Depends on: | |||
Bug Blocks: | 506276 | ||
Attachments: | build log |
Description
boxcars
2019-05-25 16:02:16 UTC
I have dev-libs/gmp-6.1.2 installed, with USE flags: asm cxx -doc -static-libs ABI_MIPS="-n32 -n64 -o32" ABI_RISCV="-lp64 -lp64d" ABI_S390="-32 -64" ABI_X86="64 -32 -x32" $ equery f dev-libs/gmp | grep libgmp.so | xargs file /usr/lib64/libgmp.so: symbolic link to libgmp.so.10.3.2 /usr/lib64/libgmp.so.10: symbolic link to libgmp.so.10.3.2 /usr/lib64/libgmp.so.10.3.2: ELF 64-bit LSB shared object, x86-64, version 1 (SYSV), dynamically linked, stripped I've filed an upstream bug: https://bugs.kde.org/show_bug.cgi?id=408131 Given that ccmake correctly finds libgmp, is it possible that there is something wrong with either how ccmake sets up $libdir stuff in the ninja files, or how ninja uses that info? Please rebuild libalkimia. (In reply to Andreas Sturmlechner from comment #3) Thanks. After rebuilding libalkimia (and a few more of kmymoney's dependencies which I had depcleaned), kmymoney builds fine. I had forgotten about this but - but just tried to compile KMyMoney from the 5.0 branch git head, and got ninja: error: '/usr/lib/libmpir.so', needed by 'lib/libkmm_mymoney.so.5.0.4', missing and no known rule to make it. mpir seems fine, with the .h file in /usr/include and all the .so files in /usr/lib64. Re-emerging mpir makes no difference, so I still think there is a problem with either cmake or ninja, or perhaps (less likely?) how kmymoney uses either of them. Should the bug title be changed to reflect this new failure, or be made more generic? Also, I updated the upstream bug with this info. Just rebuild libalkimia. This is simply the result of the profile switch. What confuses me a bit is why recompiling libalkimia helps a problem not finding libmpir.so. Is it just that it is really alkimia which pulls in mpir, and not kmymoney itself? libalkimia has libmpir.so in its INTERFACE_LINK_LIBRARIES, and when you built it under profile 17.0 the path to it stored by cmake was /usr/lib/libmpir.so when it should have been /usr/lib64/libmpir.so. It seems to be a cmake issue (find_library behaviour). *** Bug 692296 has been marked as a duplicate of this bug. *** *** Bug 699634 has been marked as a duplicate of this bug. *** It is probably time to consolidate this with the parent bug. *** This bug has been marked as a duplicate of bug 644538 *** |