Re: Threads, glibc, and all that

[hjl at lucon dot org (H.J. Lu)]

> 
> 
> I don't see closure on this issue yet, but I think I have something
> to contribute :-)
> 
>   In message <r2ogzvo4s0.fsf@happy.cygnus.com>you write:
>   > Well, the patch is ok.  But the hack shouldn't be installed
>   > unconditionally (only if the problem is real)
> So can we write a test for this?  Remember, we don't have autoconf
> to help us.

That is not a problem.

> 
>   > and how to remove it the next time?
> If we go with that patch, then I'd recommend we only install it
> in the egcs-1.0.x branch -- we fix the mainline tree in the
> "right" way, whatever that is.

We use the same test to see if it is needed or not.

> 
> Thus the band-aid's lifespan is only as long as egcs-1.0.x's life
> span.
> 
>   > This is an isolated problem which might for now be fixed with a
>   > comment on the problem in the README.
> That's another possibility.
> 
> Again, I'm leaving the decision in the hands of Jason, HJ, Ulrich
> and Mark if he wants to get involved.  I'm not qualified to render
> a decision based on the pure technical merits of any proposed patch.
> 

I will send in a patch in a day or 2.


-- 
H.J. Lu (hjl@gnu.org)