| Summary: | sys-devel/gcc-config-1.5-r1: bug in postinst() ? | ||
|---|---|---|---|
| Product: | Gentoo Linux | Reporter: | Jeremy Olexa (darkside) (RETIRED) <darkside> |
| Component: | [OLD] Core system | Assignee: | Gentoo Toolchain Maintainers <toolchain> |
| Status: | RESOLVED INVALID | ||
| Severity: | normal | CC: | prefix |
| Priority: | Normal | ||
| Version: | unspecified | ||
| Hardware: | All | ||
| OS: | Linux | ||
| Whiteboard: | |||
| Package list: | Runtime testing required: | --- | |
Not exactly related, but I did spend too much time on figuring out why it worked for some and not for some others. The end result was a changed file without a revbump for 1.4.1... Recommendation: Revbump the ebuild and mark it straight to stable so everyone has the same file. older versions shouldnt matter. the code is in postinst which means it can only be executed once you've installed a newer version. thus no revbump/stable is necessary, and actually emerging the version in question works for everyone. |
In 1.4.1, the postinst code does: if gcc-config --get-current-profile returns 0; then "do stuff" & run gcc-config $(gcc-config --get-current-profile); fi Result: gcc-config <CHOST>-<VER> In 1.5, the postinst code does: for x in $(gcc-config -C -l 2>/dev/null | awk '$NF == "*" { print $2 }'); do gcc-config $x; done Result: noop The bug is that $(gcc-config -C -l 2>/dev/null | awk '$NF == "*" { print $2 }') returns null. I am unsure of the intent, but I think someone was trying to do this: % gcc-config -C -l| grep \* | awk '{ print $2 }' x86_64-pc-linux-gnu-4.5.2 vs. % gcc-config -C -l| awk '$NF == "*" { print $2 }' % Am I wrong or misinformed? Thanks.