On Thu, 4 Dec 2003, Johnny Billquist wrote:
> On Thu, 4 Dec 2003, Brian Chase wrote:

> > Mine too :-| Looking at the top level Makefile, I gathered that one way
> > to get around this problem is to rename or remove the regress dir.  I'm
> > in the middle of rebuilding with regress renamed to regress.skip.  So
> > far so good.
>
> Yes, that is a possible solution. I wonder wether the stage for generating
> the tar-balls will be happy, though?

I'll let you know if my build gets there first :-)

> > Are you saying that the gcc in -current is broken and/or generates
> > broken code for the VAX?  That would suck.
>
> Yes. The upgrade was only a couple of days ago, and a mail was sent about
> it.

We could always build from the source of the day /before/ gcc was
broken.  I've been updating my sources between build crashes, so my
sources are only about a day old (and broken, I'd gather).  That doesn't
get us to running -current, but it gets us somewhat closer.

I must've missed the mail about the latest gcc on VAX.  What exactly is
broken with the newer compiler, and who is working to fix it?  I can
offer testing and troubleshooting cycles, if that's helpful.

> > I'm also pursuing the alternate path of getting -current running on an
> > i386 system for cross-compiling to VAX.  But that won't get around the
> > gcc problems if I'm understanding you correctly.
>
> Hmm. I actually already have that. But for this case, I would prefer to
> build it natively. (I'm always suspicious about cross-compiles.)

My thinking was that I'd use a cross-compiled release in order to get
myself on track with current, since compiling it native under 1.6.1 has
been so painful.  Then I can help beat the bugs out of -current before
the 2.0 release with native compiles.  Well, that's the plan.

-brian.