libstdc++ bootstrap failures on sparc-sun-solaris2.8 (analyzed)

[David Edelsohn]

>>>>> Gabriel Dos Reis writes:

Gabriel> OK. I'm convinced that you have a working solution for 3.0.x (and
Gabriel> maybe 3.1 if can't advance with Stephen's idea).  

Gabriel> Any objection?

Gabriel> |  I do not yet have a good idea
Gabriel> | of how or where to have only gen-num-limits care about 32x64.

Gabriel> Nearly everything:

	I was not referring to the methods affected by 32 vs 64.  What I
meant was other than testing for gen-num-limits failure with my original
patch, I don't know how or where to tie in knowledge that we are building
on a 32x64 and should pretend as if we are cross-compiling *ONLY* for
gen-num-limits. 

	I haven't really looked into where this fits into Stephen Webb's
patch which defers gen-num-limits until build time.

	My question is: assuming we can know that we are performing 32x64
(BIG IF), what do you want configure/Make to do instead of failing and
re-trying?

Thanks, David