libgcc c++ support bits moved

Zack Weinberg zackw@Stanford.EDU
Mon Oct 9 17:42:00 GMT 2000


On Mon, Oct 09, 2000 at 05:12:24PM -0700, Richard Henderson wrote:
> On Tue, Oct 10, 2000 at 12:39:30AM +0200, Hans-Peter Nilsson wrote:
> > Adding a hook, somehow filled in by the target makefile-fragment
> > instead of 'echo "You must find a way to make libgcc1 components
> > yourself"' would be a start.
> > 
> > Ok conceptually?
> 
> I guess.
> 
> I'm sorta wondering though why you don't just build those
> files from assembly instead of funky options to C.

I'd personally like to see the remaining libgcc1 logic expunged.  We
don't have any targets left that require libgcc1.c, right?

It occurs to me that now that libgcc's contents can't depend on
LANGUAGES, we could run mklibgcc at configure time.  Then we could
merge its contents into the master Makefile instead of having it be a
separate module.  And that would make it simple (relatively) for a
target to specify both an additional file to compile into libgcc, and
a rule to build it.

I should dust off my nonrecursive make patch, the problems with it
were mostly caused by the c++ additions to libgcc if memory serves.

zw



More information about the Gcc-patches mailing list