When you set MAKEOPTS=-j* on Gentoo/FreeBSD, following error sometimes occurs. <snip> /usr/lib/portage/python3.4/ebuild-helpers/ecompressdir: cannot make pipe for process substitution: File exists /usr/lib/portage/python3.4/ebuild-helpers/ecompressdir: line 72: /var/tmp/portage/sys-freebsd/freebsd-lib-10.3_beta1/temp/sh-np-2081406713: ambiguous redirect <snip> * The ebuild phase 'install' with pid 56569 appears to have left an orphan * process running in the background. >>> Installing (1 of 1) sys-freebsd/freebsd-lib-10.3_beta1::local_10_3 ... Workaround, # mkdir -p /etc/portage/env # echo 'CFLAGS="${CFLAGS} -DUSE_MKTEMP=1 -DUSE_MKSTEMP=1"' > /etc/portage/env/bash-mktemp-fix # mkdir -p /etc/portage/package.env # echo 'app-shells/bash bash-mktemp-fix' >> /etc/portage/package.env/bash FYI, https://svnweb.freebsd.org/ports?view=revision&revision=391614 https://svnweb.freebsd.org/ports?view=revision&revision=391692 Reproducible: Always
Created attachment 425192 [details, diff] sample patch for bash-4.3_p42-r2.ebuild
CC'ing base-system. @base-system: Any issues with the proposed patch? This issue happens a fair bit on an SMP Gentoo/FreeBSD system and passing MAKEOPTS with a high -j.
i don't see why we wouldn't define these for all targets. i don't think it needs to be restricted to FreeBSD. i've posted upstream. let's see what they have to say.
upstream doesn't have a reason for defaulting off, so pushed for all versions: https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=9215aba15eba99f7284abb90e9fde8d76c56db1c i'll do a revbump in a follow up as there's one or two bash bugs pending.