parallel builds revisited

[youshi10 at u.washington.edu]

On Thu, 12 Apr 2007, Benjamin Lutz wrote:

> On Thursday 12 April 2007 22:20, Pav Lucistnik wrote:
>> Benjamin Lutz píše v čt 12. 04. 2007 v 21:07 +0200:
>>> On Thursday 12 April 2007 11:06, Garrett Cooper wrote:
>>> > I dunno how you want to approach this, but gmake does recommend 2
>>> > jobs be run in parallel for HTT enabled chips, and 3 or 4 jobs
>>> > for a dual core machines.
>>> > -Garrett
>>>
>>> So far the approach is one job per CPU. I'll do some benchmarks
>>> lateron to determine wether it really helps to run more jobs. For
>>> the KDE ports,
>>
>> Sadly, Qt blows up with -j2, so here comes KDE hopes...
>
> Must be a result of their qmake generated makefiles. Fortunately, moth
> software using autotools (kde*, and also xorg7) seems to build just
> fine with -j.
>
> A bigger blow than Qt is that OOo doesn't work with -j though :( That's
> probably the port that would need it the most... (the failure seems to
> happen very late in the build process though, which gives me hope that
> it's something reasonably easy to fix. If only the development cycles
> (code, build, test) weren't so extremely long...)
>
> Cheers
> Benjamin

Their make system needs to be seriously revised though, as well as their source base IMHO.

They're still using Mozilla from 2k3 (1.7b) instead of contemporary Mozilla browsers by default, I noticed. It's a mess how everything's done..

OOo could be mapped out as modular components though, instead of a straight line of dependencies, and once there are no longer any core dependencies, the make jobs can diverge building things like Calc, Graph, Writer, etc.

This requires a threaded or fork / waitpid based gmake though that doesn't currently exist AFAIK.

-Garrett