Patch: java.io.PrintStream

[Michael Koch]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Am Sonntag, 8. Juni 2003 15:55 schrieb Mark Wielaard:
> Hi,
>
> On Sun, 2003-06-08 at 12:48, Michael Koch wrote:
> > I would like to merge java.io.PrintStream from classpath. This
> > basically uses a PrintWriter object internally instead of doing
> > it all alone.
> >
> > Can somebody with working Mauve check if this gives any
> > regression ?
>
> Yes, there is the following failure:
> FAIL: gnu.testlet.java.io.PrintStream.subclass (number 1)
>
> That test tries to set the protected out variable (of the
> FilterOutputStream superclass of PrintStream) and then use it
> normally. Which fails with this new PrintWriter delegate thing.
> Tricky to get right...

Hmm, will have to look into this test later ...

> > I somehow can get Mauve working here.
>
> What is the problem? I just checked in a little fix to make mauve
> work with gcj (native standalone) again.

I try to use mauve as part of the libgcj testsuite. The problem is 
that whatever I try compiling the tests work but linking doesnt. I 
investigated that several times and it seems that the linker is 
called in the wrong directory or on directory depth is ditched when 
giving filenames to the linker.

> > Comments ?
>
> I would loose the synchronized on the write methods. The original
> libgcj implementation didn't have them and I don't believe the
> PrintStreams are guaranteed to work deterministically when used
> from multiple Threads.

Yes, you are right. I will remove the "synchronized" keywords.


Michael
- -- 
Homepage: http://www.worldforge.org/
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.1 (GNU/Linux)

iD8DBQE+41HJWSOgCCdjSDsRAikbAJ4v4eAM+bpLjL6nJndg7p6CX35fBQCeJH3H
slYSwWdd9jsG4Wg7Jl/MgIY=
=AbQ7
-----END PGP SIGNATURE-----