| Summary: | fortran.eclass forbids FC=${CHOST}-gfortran | ||
|---|---|---|---|
| Product: | Gentoo Linux | Reporter: | Nathan Phillip Brink (binki) (RETIRED) <binki> |
| Component: | Eclasses | Assignee: | Kacper Kowalik (Xarthisius) (RETIRED) <xarthisius> |
| Status: | RESOLVED OBSOLETE | ||
| Severity: | normal | CC: | esigra, kugelfang, sci |
| Priority: | High | ||
| Version: | unspecified | ||
| Hardware: | All | ||
| OS: | Linux | ||
| Whiteboard: | |||
| Package list: | Runtime testing required: | --- | |
| Bug Depends on: | |||
| Bug Blocks: | 306835, 339465 | ||
| Attachments: |
build.log
die.env emerge --info allows programs to compile if FC=$(portageq envvar CHOST)-gfortran |
||
Created attachment 234939 [details]
build.log
Created attachment 234941 [details]
die.env
Created attachment 234943 [details]
emerge --info
Created attachment 238971 [details, diff]
allows programs to compile if FC=$(portageq envvar CHOST)-gfortran
This patch causes the ebuilds not to die just because FC=$(portageq envvar CHOST)-gfortran.
Some packages, like xfoil, test ${FOTRANC} = "gfortran" instead of using fortran_patch(). I couldn't find a test-case for my modified fortran_patch() in the time I gave myself.
fortran.eclass not used anymore in the tree |
Reproducable on portage with $ FC=$(portageq envvar CHOST)-gfortran emerge -v blas-atlas Would also cause problems with the -gfortran that tc-getPROG() returns. ${CHOST}-gfortran is valid and must be supported by the eclass. Perhaps the eclass can automatically strip ${CHOST} from FC or just ignore everything before and including the last `-' in FC.