glibc and egcs....
Christopher C Chimelis writes:
>
> Hi all...
>
> Ok, after re-reading the exchange about the __register_frame_info problem,
> I made a few decisions...
>
> Right now, egcs won't build due to a couple of things (a makeinfo problem
> which I've solved and a Makefile problem which I'm still working on).
> Glibc, however, will. BUT, because we need to remove the HJL patch that
> creates the __register_frame_info problem to begin with, I'm holding off
> updating either until we can resolve the egcs packaging problems.
>
> I just wrote to the egcs maintainer and, given his past track record, he
> should be able to solve this by the end of the week (he's very prompt).
> After I get a working copy of egcs and run a test or two, I'll go ahead an
> build the latest glibc sans __register_frame_info.
I was thinking there was a consensus glibc must contain
register_frame_info, as opposed to the crtbegin.o that should not
reference it (or if it reference it, it must contains a weak
definition as I proposed originally)
so:
- programs build even after now can run on any system (whereas glibc
has the symbol or not).
- programs that for some reason depends on it (such as all that are built
with our current egcs) can run on an uptodate system..
Loic
Reply to: