Re: More Objective C @"string" bugs...

["Zack Weinberg" <zackw at stanford dot edu>]

On Tue, Mar 06, 2001 at 09:53:22AM -0700, Tom Tromey wrote:
> >>>>> "Joseph" == Joseph S Myers <jsm28@cam.ac.uk> writes:
> 
> Joseph> GCC developers follow the standard procedure of running the
> Joseph> testsuite and avoiding regressions - it's simply that the ObjC
> Joseph> testsuite is very small so problems are likely not to be shown
> Joseph> up.
> 
> GCC developers should do this, but they don't.  The Java compiler
> still randomly breaks in easily-noticeable ways (eg, can't build
> libjava).  For instance it is broken now.  The real state of affairs
> for GCC is that if C and C++ work, then the patch goes in.

I can't speak for anyone else, but I personally have *never* been able
to build libjava.  The "size of <symbol>" error is just the most
recent issue.  For instance, I still have the problem with the
libstdc++ -I switches disappearing, that was being discussed last week.

I try it periodically, but in order to get anything done I have to
turn the Java front end off.  I wouldn't be surprised if other people
have the same problem.  Once Java stabilizes to the point where it can
be reliably built, I expect people will leave it on, and will not
break it so much.  Yes, catch-22.

zw