shared libstdc++ is broken on hppa1.1-hp-hpux9.01

[Richard Henderson]

On Sat, Jul 24, 1999 at 10:26:45AM -0600, Jeffrey A Law wrote:
>   In message < 19990721164254.D19378@cygnus.com >you write:
>   > It's better to define a nonlocal_goto expander than to punish
>   > local indirect jumps with such a sequence.
> Agreed.

I should correct myself here -- define a longjmp expander.  

Non-local gotos are guaranteed to jump to another function in the same
unit of translation.  That is to say, the concept not the expander since
we'll fall back on nonlocal_goto when implementing longjmp since on 
many platforms they have the same requirements.


r~