Re: [RFC] unify dynamic_symbol_p implementations

[Alan Modra <amodra at bigpond dot net dot au>]

On Mon, Jul 21, 2003 at 04:25:34PM -0700, Ulrich Drepper wrote:
> H. J. Lu wrote:
> > There were different interpretations of how weak symbols should work
> > and we discussed it at the ia64 psABI meeting. The conclusion was
> > glibc was wrong.
> 
> glibc was not wrong (please is impossible given the behavior is
> unspecified).  But coming to a common, reasonable semantics made more
> sense with glibc's behavior changing (Irix would be effected, too, if
> they would make such a change).

Has someone written up a draft psABI that defines weak symbol behaviour?
Or better, a draft gABI?  gABI on www.caldera.com is still 24 April 2001.

Also, if ld follows the new semantics, which versions of glibc will no
longer build?  Not that this should deter us from changing ld, but we'd
need to mention in release notes that older versions of glibc require
older versions of binutils.

-- 
Alan Modra
IBM OzLabs - Linux Technology Centre