compatibility with jikes ?
Fri Feb 28 10:18:00 GMT 2003
-----BEGIN PGP SIGNED MESSAGE-----
Am Donnerstag, 27. Februar 2003 17:25 schrieb Tom Tromey:
> >>>>> "Michael" == Michael Koch <email@example.com> writes:
> Michael> java.net.DatagramSocketImpl.getOption()
> Michael> java.net.DatagramSocketImpl.setOption()
> Michael> java.net.SocketImpl.getOption()
> Michael> java.net.SocketImpl.setOption()
> Michael> They came in via a merge with classpath long time ago. GCJ
> Michael> dont need them. They are only there to make these classes
> Michael> compilable with jikes (in classpath tree).
> Michael> Should we really keep them in the libgcj tree for jikes
> Michael> compatibility or better remove them because we dont need
> them Michael> ?
> Actually the comment says they are there to work around a gcj bug.
> This may be one of those bugs that only shows up if you compile a
> lot of java files with one gcj command. Though it does look like a
> classic instance of the Miranda method problem, so I'm not entirely
> sure why we wouldn't need this.
> Generally I prefer to keep things merged with Classpath. That
> makes it simpler to know when we need to import a bug fix or
> OTOH, java.net has already diverged quite a bit. We're already
> going to have to spend time re-merging. So if you want to do this,
> it is fine by me.
I'm planing this. Some work is done locally already. Will submit
patches part by part.
> What I'd really like is a fix for this bug in gcj.
I tried compiling classpath with
gcj 3.2.3 20030221 (Debian prerelease).
With jikes and trunk all is okay. With 3.2.3 I got the following
/usr/bin/gcj-3.2 --bootclasspath '' -classpath
- -C -d . @classes
Superclass `AttributedCharacterIterator.Attribute' of class
`java.awt.font.TextAttribute' not found.
public final class TextAttribute extends
At least this means the bug is fixed in trunk. I couldnt test with 3.3
I hope this helps a little bit.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.1 (GNU/Linux)
-----END PGP SIGNATURE-----
More information about the Java