Today I ran distccmon-gnome on one of the slow ones to see how distcc behaved and saw some big packages (like metacity) not being distributed to other machines.
Looking at the ebuild I found
Code: Select all
MAKEOPTS="${MAKEOPTS} -j1"Is this due to bad Makefiles or is the example above the ONLY workaround?
If it is not: why do we allow us this workaround? Would it not be better to bug upstream to get their Makefiles right or patch it?
I think the above example is the wrong way to go since it gives problems with distcc, SMPs will not have their full potentials used and last but don't least: sometimes this "fix" is kept even when the problem is fixed upstream.
For example metacity: in the bugtracker there is in the bugreport for the issue of having parallel buildings at least one comment from a person who did not have problems compiling metacity-2.6.* (a package from 2003) with parallel processes. On my machine a edited ebuild (2.16.* and 2.17.*) worked with distcc and MAKEOPTS="-j" and that spawned an AWFUL lot of processes threaded over at least three diffrent machines and it compiled fine.
So is it time to place a QA for
Code: Select all
MAKEOPTS="${MAKEOPTS} -j1"
