This is the mail archive of the
libstdc++@gcc.gnu.org
mailing list for the libstdc++ project.
Re: probs cross-building current libstdc++-v3, because of need to link?
- From: Daniel Jacobowitz <drow at mvista dot com>
- To: Phil Edwards <phil at jaj dot com>
- Cc: cgd at broadcom dot com, gcc at gcc dot gnu dot org, libstdc++ at gcc dot gnu dot org
- Date: Thu, 14 Aug 2003 12:52:58 -0400
- Subject: Re: probs cross-building current libstdc++-v3, because of need to link?
- References: <yov5bruvmmr8.fsf@ldt-sj3-010.sj.broadcom.com> <20030812063122.GB7772@disaster.jaj.com> <20030812064732.GA8501@disaster.jaj.com>
On Tue, Aug 12, 2003 at 02:47:32AM -0400, Phil Edwards wrote:
>
> My hands wrote the following before consulting my brain:
>
> > ...yah, methinks I know what I've forgotten. Patch in progress.
>
> Belay that. Patch not in progress. Sleep in progress.
>
> There is a semi-nifty macro in modern autoconf called AC_NO_EXECUTABLES,
> which turns off the tests involving a linker. But it can't be called
> unconditionally, because in a native build we want to do those link tests.
> Nor can it be called conditionally, like
>
> if (it's a cross build) then
> AC_NO_EXECUTABLES
> fi
>
> because it's not just shell code; it has other effects at 'autoconf' time.
>
> In fact, trying to use it at all causes fatal autoconf errors. (One of
> the "other effects" is to flag an error if you try to do a linking test.
> Which we do. They're not on the same code path, but autoconf can't
> know that.)
>
> We may have to do something devious here, such as redefining the same
> things that AC_NO_EXECUTABLES does, but without the killer side effects,
> but right nBrain panic: falling asleep in 10 secoNO CARRIER
I posted just the macro you're looking for to autoconf-conversion a
month or two ago. Still have a copy of it?
--
Daniel Jacobowitz
MontaVista Software Debian GNU/Linux Developer