On wto, 2017-07-04 at 23:33 +0200, Manuel Rüger wrote: > On 04.07.2017 21:29, Zac Medico wrote: > > On Fri, Jun 30, 2017 at 2:49 AM, Manuel Rüger <mr...@gentoo.org> wrote: > > > + > > > COMPRESSION_COMMAND=$(PYTHONPATH=${PORTAGE_PYTHONPATH:-${PORTAGE_PYM_PATH}} > > > \ > > > + "${PORTAGE_PYTHON:-/usr/bin/python}" > > > "$PORTAGE_BIN_PATH"/binpkg-helper.py \ > > > + compressioncmd ${CATEGORY}/${P}) > > > + [ -z "${COMPRESSION_COMMAND}" ] && \ > > > + die "Failed to get COMPRESSION_COMMAND" > > > tar $tar_options -cf - $PORTAGE_BINPKG_TAR_OPTS -C "${PROOT}" . | > > > \ > > > - $PORTAGE_BZIP2_COMMAND -c > "$PORTAGE_BINPKG_TMPFILE" > > > + $COMPRESSION_COMMAND -c > "$PORTAGE_BINPKG_TMPFILE" > > > assert "failed to pack binary package: '$PORTAGE_BINPKG_TMPFILE'" > > > PYTHONPATH=${PORTAGE_PYTHONPATH:-${PORTAGE_PYM_PATH}} \ > > > "${PORTAGE_PYTHON:-/usr/bin/python}" > > > > If all that we really need is COMPRESSION_COMMAND, then the helper > > script is overkill. We should just pass a variable from > > doebuild_environment function. The variable name must be prefixed with > > PORTAGE_. > > > > Also, note that your mail client wrapped lines in this patch. > > > > Thanks for the review Zac! > > We'd need a bit more as the COMPRESSION_COMMAND depends on > ${CATEGORY}/${P} here in order to avoid a catch22 when using a > decompressor that is set to something a standard install doesn't include. > > Assume all binpkgs are set to be compressed with zstd, the patch makes > sure an app-arch/zstd binpkg will still be compressed with bzip2.
And app-arch/bzip2 binpackage? In other words, that's the wrong approach. If I set binpackages to use X, I want them to use X, not 'use X if someone thought it correct to use X'. Just do what the user says to do. If you try to outsmart him, you're building another Windows. -- Best regards, Michał Górny
signature.asc
Description: This is a digitally signed message part