[ANNOUNCE] Libgcj in gcc tree availability.

[Mo DeJong]

On Tue, 12 Dec 2000, Bryce McKinlay wrote:

> Mo DeJong wrote:
> 
> > Don't you need to run autoconf after you edit configure.in,
> > and does autoconf need to be the Cygnus maintainer tools version?
> > I tried re-running with autoconf 2.13 and it did not work. I would
> > think it would be easier to have a --enable-libgcj flag
> > than forcing people to hose around in the configure.in file.
> > Is that possible?
> 
> 
> The top level configure.in doesn't actually have anything to do with 
> autoconf (Its just named that way to confuse you ;-)
> 
> So, no, you don't need to run autoconf.

Thanks, I have now got a working gcj build! I ran it though
the Jacks test suite and everything looks good (no regressions).

The one thing I found that really bothers me is the following:

% cat T1410d1.java
class T1410d1 {
    void foo(int i) {
        switch (i) {
            case 0:
            //case 1:
            case 0:
        }
    }
}

If you compile this class as-is, it compiles (incorrectly I might add).

If you uncomment the "case 1:" the compiler will core dump.
Should I file a PR on this or is it already a known problem?

Both types should fail to compile because of a duplicate case label.

I poked around in gnats but I did not see anything.

thanks
Mo DeJong
Red Hat Inc