Re: `make check' overworks one core on my Core2 quad

[Graham Percival]

On Sun, Dec 13, 2009 at 02:27:53PM -0200, Han-Wen Nienhuys wrote:
> On Sun, Dec 13, 2009 at 2:02 PM, Graham Percival
> <address@hidden> wrote:
> >> For now, the easy fix is to use -j1 with CPU_COUNT for building the docs.
> >
> > Yes; that became the recommended way in Oct.  But it's much slower
> > than it would otherwise be.
> 
> That surprises me. I thought the majority of the time was spent
> running LilyPond, which would be efficient with CPU_COUNT set, so the
> difference should be small.

Ack!  Sorry, I misread your solution.  It never occurred to me to
try -j1 CPU_COUNT=4.  I'll do that tomorrow.

> >> A more elaborate solution would be either some kind of locking, or to
> >> check whether the .ps / .pdf exists before actually processing the
> >> .ly; the latter is still suscepitible to races, though, but a check
> >> could make the opportunity window smaller.
> >
> > I think a .lock would be good.  That's the typical solution to
> > everything when it comes to parallel processing.  :)
> 
> Right - the script should do an flock() on the database directory when
> opening it.
> 
> (doesnt work on NFS though)

On the entire directory, or just on the lybook-db/??/ dir?

Cheers,
- Graham