On Sat, 9 Mar 2013 18:50:36 +0100 Michał Górny wrote: > On Sat, 09 Mar 2013 19:26:34 +0200 > Samuli Suominen wrote: > > > I have to run off for few hours, like now, but this one user is filing > > dozens of these epunt_cxx bugs and nobody seem to be able to reproduce them: > > > > http://tinyurl.com/bjwxzad > > > > Could someone verify they are actually legimate before they get assigned > > to maintainers? > > Oh, I see the problem now. Stupid bash, stupid loops and stupid > subshelling. I'll attach a patch for the check in a minute. Index: eutils.eclass =================================================================== RCS file: /var/cvsroot/gentoo-x86/eclass/eutils.eclass,v retrieving revision 1.411 diff -u -B -r1.411 eutils.eclass --- eutils.eclass 26 Feb 2013 14:36:40 -0000 1.411 +++ eutils.eclass 9 Mar 2013 17:54:20 -0000 @@ -1270,11 +1270,11 @@ [[ -z ${dir} ]] && dir=${S} ebegin "Removing useless C++ checks" local f any_found - find "${dir}" -name configure | while read f ; do + while IFS= read -r -d '' f; do patch --no-backup-if-mismatch -p0 "${f}" \ "${PORTDIR}/eclass/ELT-patches/nocxx/nocxx.patch" > /dev/null \ && any_found=1 - done + done < <(find "${dir}" -name configure -print0) if [[ -z ${any_found} ]]; then eqawarn "epunt_cxx called unnecessarily (no C++ checks to punt)." ----- But that doesn't seem to be the issue. If you remove '>/dev/null', you'd notice that the patch does not apply... -- Best regards, Michał Górny