Summary: | recent libgfortran-related commits breaks cantor-R support on fedora | ||
---|---|---|---|
Product: | [Applications] cantor | Reporter: | Rex Dieter <rdieter> |
Component: | general | Assignee: | Alexander Rieder <alexanderrieder> |
Status: | RESOLVED FIXED | ||
Severity: | normal | CC: | aacid, kevin.kofler |
Priority: | NOR | ||
Version: | unspecified | ||
Target Milestone: | --- | ||
Platform: | Other | ||
OS: | Linux | ||
Latest Commit: | http://commits.kde.org/cantor/aee382880f4f7a3b8d02bf0b697e6b334eacf65a | Version Fixed In: | 4.10.1 |
Sentry Crash Report: |
Description
Rex Dieter
2013-02-01 14:43:58 UTC
Fwiw, I question the legitimacy of the comment in FindR.cmake: # for at least some versions of R, we seem to have to link against -lRlapack. Else loading some # R packages will fail due to unresolved symbols, or we can't link against -lR. I'd argue the problem more likely lies in some broken R installs, where their modules/libraries themselves are underlinked and/or missing symbols, and this here is just a workaround/bandaid. (I suppose that may discount static linking though... ugh. Not sure if that's a case worrying to much about though honestly) Rex, does https://git.reviewboard.kde.org/r/108701/ help? Looks like it will, I'll test and report back. Any luck? reviewboard patch confirmed good fixing this bug, thanks. Git commit aee382880f4f7a3b8d02bf0b697e6b334eacf65a by Christophe Giboudeaux. Committed on 07/02/2013 at 23:06. Pushed by cgiboudeaux into branch 'KDE/4.10'. Try harder to find gfortran Some distributions (Fedora, OpenSUSE...) don't install the libgfortran library in a standard dir thus making the detection fail. Instead of looking for libgfortran.so, we know run gfortran to get the full path. REVIEW: 108701 FIXED-IN: 4.10.1 M +16 -3 cmake/FindR.cmake http://commits.kde.org/cantor/aee382880f4f7a3b8d02bf0b697e6b334eacf65a |