David Boyce wrote:

> On Wed, May 4, 2011 at 11:20 AM, Per Jessen <[email protected]> wrote:
>> yeah, SYNCSH_SERIALIZE is equal to what I do today using this
>> construct:
>>
>> (flock -s 200; some-command $^ $@) 200>/var/lock/some-lockfile
>>
>> It satisfies the single-thread requirement, but in massively parallel
>> runs, several of these often end up waiting for each other. It would
>> be optimal if make knew not to submit more than 1 at a time, I was
>> just wondering if I'd skipped a page in the manual :-)
> 
> Could you rephrase this? I can't see a meaningful distinction between
> "waiting for each other" and "submit one at a time". Either way
> they're serialized, no?

Yes, it satisfies the serialization requirement, but if, for instance,
you run a make -j16 and 3 processes get tied up waiting for a fourth,
the jobserver thinks they're running and won't be submitting jobs that
could run. It's purely an optimization issue.


/Per Jessen, Zürich


_______________________________________________
Help-make mailing list
[email protected]
https://lists.gnu.org/mailman/listinfo/help-make

Reply via email to