[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Gcl-devel] gcl-2.6.2 / 303 ansitests failed

From: Camm Maguire
Subject: Re: [Gcl-devel] gcl-2.6.2 / 303 ansitests failed
Date: 20 Jul 2004 18:11:49 -0400
User-agent: Gnus/5.09 (Gnus v5.9.0) Emacs/21.2


"Mike Thomas" <address@hidden> writes:

> Hi Camm.
> | Just to double check, you know about --enable-dynsysgmp, yes?  Hope it
> | works on mingw.
> Yes; I hadn't used it until now due to a concern for keeping builds
> standardised and lack of understanding that the patched mul_n is used
> whether an external lib is used or not.
> Rather than use a dynamic library I used a precompiled static lib and
> everthing built.  Whether it works remains to be seen as I am out of time.

I am actually surprised in linked.  Shared libs, unlike static libs,
are supposed to have 'weak' symbols which can be overridden by 'strong'
ones.  In the static case I typically get a multiple definition error.
-u forces the symbol to be undefined at the beginning, and therefore
pulled in from its version in libgcl.a, which precedes libgmp on the
linker command line.  (Otherwise, the symbol is only undefined where
it is directly used, within libgmp itself, which then is where the
definition is found, as the linker proceeds on a fifo basis.)

> Strangely it went through a complete GMP configure - the configure phase
> took about 5 minutes against the build itself - 7 minutes.
> If gmp is built we're talking about 25-30 minutes for a complete build so as
> you can see this is a major saving for me.

Yes, we can improve this further.  Please don't let me forget.

Take care,

> Cheers
> Mike Thomas.

Camm Maguire                                            address@hidden
"The earth is but one country, and mankind its citizens."  --  Baha'u'llah

reply via email to

[Prev in Thread] Current Thread [Next in Thread]