cvs commit: ports/lang/q Makefile

[Wesley Shields]

On Mon, Apr 13, 2009 at 07:38:22PM +0300, Ion-Mihai Tetcu wrote:
> On Mon, 13 Apr 2009 10:04:29 -0400
> Wesley Shields <wxs at FreeBSD.org> wrote:
> 
> > On Mon, Apr 13, 2009 at 04:43:06PM +0300, Ion-Mihai Tetcu wrote:
> > > On Mon, 13 Apr 2009 08:42:52 -0400
> > > Wesley Shields <wxs at FreeBSD.org> wrote:
> > > 
> > > > On Mon, Apr 13, 2009 at 04:08:42AM -0700, QAT at FreeBSD.org wrote:
> > > > > The Restless Daemon identified a PLIST error while trying to
> > > > > build: q-7.11 maintained by gahr at FreeBSD.org
> > > > >  Makefile ident: $FreeBSD: ports/lang/q/Makefile,v 1.35
> > > > > 2009/04/13 09:53:27 gahr Exp $
> > > > > 
> > > > > Excerpt from
> > > > > http://QAT.TecNik93.com/logs/7-STABLE-FPT-NPD/q-7.11.log :
> > > > 
> > > > [...]
> > > > 
> > > > This looks like a false positive to me.
> > > 
> > > It wasn't. There was a second commit that fixed the problem probably
> > > while the port was building.
> > 
> > Ah, and by the time I got to see the log it was fixed.  Thanks.
> 
> Yup.
> 
> > > > On a related note, how does QAT handle MAKE_JOBS_SAFE?  Does it
> > > > manage to find the error which could have been much earlier in
> > > > the log?
> > > 
> > > QAT has 4 CPUs so it builds with multiple make jobs if the ports
> > > says so. There's no supplementary analyses being done, if the build
> > > doesn't exit when one of the jobs fails (as it should), the error
> > > would be catch at install or packaging time.
> > 
> > But then the log excerpt in the email may not actually contain the
> > error.
> 
> The excerpt is just a tail on the last lines of the log; usually it
> catches the error; I could try to do a little heuristics, but I'm not
> that sure of the results (I'd use the markup patters used by webui).

OK, that's what I thought.  I'm not suggesting that be done, but just
that it's a known thing (I hesitate to even call it a problem).

> Anyway, the log snippet is intended to hint the problem and any
> responsible commit implies reviewing the hole log.

Absolutely.

-- WXS