Re: GCC Make Check

[Jeff Bailey]

On Fri, May 10, 2002 at 05:42:19PM +0200, Marcus Brinkmann wrote:

> > I remember you mentioning on IRC that `make check' for gcc won't
> > work on the Hurd.  Was that a stdio-related problem, or other?

> If it is what I have in mind, it is a not having ulimit that works
> correctly type of problem (eg test trying to allocate infinite
> amount of memory)

Hrmm.  Is that a problem only when running as root, or does it even
exist as a user?

> > Having *just* caught a MAXPATHLEN bug in gcc-3.1 (ada front end),

> Eeek!

Yeah.  It's a particularily nasty one, in that the ada front end
requires an existing ada compiler to run.  So until now, there's no
way we could've caught it.

> > I'd like to try and implement a compile farm of some sort so we
> > can catch these problems in major GNU packages very soon after
> > they're checked into CVS.  I just need to see if I can find
> > another machine here (I think I can, or if we can solve the
> > crashing problem,

> crashing problem?  what crashing problem?  someone just needs to reboot it!

Sure - But they need to reboot it because it crashed, right? =)

> > perhaps we can use a machine at the FSF - I'm sure we won't have
> > trouble getting one donated or assigned).

> > It would be nice in those cases to run `make check' as well as
> > compiling.

> Do you think it is worth to work on a CVS module for turtle?

Yes.  If you have time for the task, great, if not I was hoping to
solicit volunteers after writing a spec for it.  There seem to be a
few bored people hanging out on #hurd who might be willing to take
this on.

If you don't have time, I was going to look at Tinderbox, but I've
heard it particularily nasty to setup.

> BTW, Jeroen might be interested in monitoring the build logs etc.

That would be cool.  But I'll let him volunteer before I give him
work. ;)

-- 
 One of the great things about books is sometimes
 there are some fantastic pictures.
 -- George W. Bush