Code review comment for lp:~jelmer/bzr-builddeb/parallel-test

On Sun, Feb 27, 2011 at 11:32 AM, James Westby
<email address hidden> wrote:
> On Sat, 26 Feb 2011 22:17:08 -0000, Robert Collins <email address hidden> wrote:
>> On Sun, Feb 27, 2011 at 11:07 AM, Jelmer Vernooij <email address hidden> wrote:
>> > Hmm, perhaps we should force it to BZR_CONCURRENCY=1 if parallel isn't in DEB_BUILD_OPTIONS?
>>
>> why? That would be slower than just not passing --parallel in. And
>> unlike make -j, bzr won't spawn until death, it creates precisely the
>> number of workers as there are cpus.
>
> Well, don't pass --parallel in that case then.
>
> I would think that the implied meaning of not passing parallel=n would
> be parallel=1.

I don't recall debian policy saying anything specific. Anyhow, as I
say, bzr handles default parallelism much better than make and many
other similar tools.

-Rob