[distcc] Knock, Knock, can distcc come out to play?

[Martin Pool]

On 12/09/05, John Carter <john.carter at tait.co.nz> wrote:

> Until someone switches off his machine in the middle of someones elses
> compile.

If you just kill off the main distcc process but leave the children
running then they'll exit when that job is finished.  This is a nicer
way to shut down.  I'm not sure exactly how this would be accomplished
on Windows but perhaps it's possible.

distcc could be improved by having the main process wait until all the
jobs are complete before it finally exits.

One the one hand, a machine that suddenly disappears in the middle of
a job is potentially a problem and ought to be reported.    On the
other hand if it's just a machine entering or leaving it shouldn't
produce such large errors.

> Is there anyway of just getting distcc to just quietly drop that IP out of
> the farm and carry on without saying anything?

I think the behaviour is OK, and it's just the warning messages that
are a problem.  It could fairly easily be changed to just log those
messages and not display them to stderr.

-- 
Martin