[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Gcl-devel] FreeBSD fixes to HEAD for GCL
From: |
Mark Murray |
Subject: |
Re: [Gcl-devel] FreeBSD fixes to HEAD for GCL |
Date: |
Fri, 10 Oct 2003 23:21:18 +0100 |
Camm Maguire writes:
> Greetings, Mark, and thank you *so* much for your suggestions! This
> is just a quick note to state that I'd like to send you a longer more
> detailed reply tomorrow. In short, I'd really like to get this to
> compile out of the box on xBSD, and am prepared to do the commits to
> make it happen. But there is at least one known instance in which
> gmp3 has to be patched to avoid a potential gc bug/conflict. Schelter
> put this in himself, and even when configuring with dynsysgmp, gcl
> will compile this file and use its symbols in prefece to their analogs
> in the gmp3 library via the -u command line option to ld. (see the
> creation of raw_gcl). So while I agree that having the gmp3 tree is
> bloat, we cannot entirely dispense with it. WFS asked the gmp3
> developers to include the patch, but they declined to my
> understanding. And 10000! is, unfortunately, not a reliable test at
> all in this situation.
Hiya. I had a strong suspicion that 10000! wouldn't cut it. What would
be a better test? (I'm motivated to actually fix this, now!) BTW - there
is hardware "out there" that can be used as a BIGNUM accelerator, and
this is one of the reasons I'm trying to avoid a fork.
> The BFD tree is only there for possible future patches on platforms
> where bfd relocation does not work. It appears that the first such
> case might be Mac OS X with Aurelian's changes. But for mainline
> boxes it can be elided.
Hmm. Can it not be assumed that whatever BFD is ported to OSX will do
the job?
M
--
Mark Murray
iumop ap!sdn w,I idlaH