When compilation of sci-electronics/spice-3.5.5 fails (*this* was someone's PEBKAC, so this is *not* the reason of this bug), portage installs broken package(!), because ./util/build won't exit with exit-status 1 even if it fails. I'd recommend adding something like this: || die "dobin failed" after dobin at least. Better would be fix the build-script, because it is broken, but I don't know how, so I can't help (yet). You can also inherit fixheadtails eclass and run this after epatching: ht_fix_file util/build because it uses old head/tail syntax.
(In reply to comment #0) [...] > I'd recommend adding something like this: > > || die "dobin failed" > > after dobin at least. Better would be fix the build-script, because it is > broken, but I don't know how, so I can't help (yet). Fixed, thanks. > You can also inherit fixheadtails eclass and run this after epatching: > > ht_fix_file util/build > > because it uses old head/tail syntax. Fixed too, but using a simple sed instead. This way, if the fixheadtail eclass needs to be removed some day, this package won't be an issue. Also, I have fixed some rather ugly issue with the mtime of the unpacked source directory which forced WORKDIR to be recreated each time you issue an 'ebuild <command>'. What happens is that the files that were previously built are then missing from ${WORKDIR}, but the .compiled file is still there in ${PORTAGE_TMPDIR}/portage/${PF}, thus nothing is recompiled and installation fails. This may have been the root cause of your installation failure above if you installed using step-by-step ebuild commands instead of emerge. By the way, do that 'someone' a favor, and tell him/her to use gnucap, or at the very least ngspice, instead of ancient spice. ;o) Thanks for your (interesting) bug report. Denis.