Re: 0.05

[Mark Wielaard]

Hi,

On Mon, 2002-12-16 at 22:36, Brian Jones wrote:
> * Integrate any remaining patches on Savannah as needed.

I still have some patches "claimed" but I got absorbed in gcc 3.3
testing and didn't have time to do more merging. I am currently looking
at merging/fixing the ObjectStream classes between Classpath and libgcj
but ran out of time and I probably won't have time the rest of the week
to look after the JRVM patches. So feel free to analyze and/or patch the
remaining items.

> * Fix Classpath bugs found by running Mauve with GCJ, Kissme, ORP, or SableVM.

I made a list for gcj here:
http://gcc.gnu.org/ml/java/2002-12/msg00185.html

> * Ensure support of GNU Crypto as a full security provider.

There is one remaining bug (but with patch) see:
http://gcc.gnu.org/ml/java-patches/2002-q4/msg00430.html
But that should be all there is to it. I wrote a couple of Mauve tests
that ensure our security provider framework wroks as expected. And GNU
Crypto now contains a couple of tests to make sure that it is
functioning correctly that (with the above test) now all succeed (at
least with gcj, Kissme was so slow that I didn't test it properly).

> Comments on this list, suggestions for additions or removals, etc,
> welcome.

List looks fine. I would concentrate on the pending patches and open
bugs. Extra packages are nice to have but not that

> I think our release criteria should be:
> 
> * Work with Kissme and JRVM, SableVM, and ORP if possible from the release.

Kissme from CVS works fine with the current Classpath from CVS. But I
haven't tested any others (except gcj of course). You might want to take
as a criteria that it should buildwith the current (3.2) and next (3.3)
release of gcj and the latest jikes release (which it should already).

> * Full javadoc generated by gjdoc available for general consumption.

There is only one open issue here that I know of. The produced HTML
pages should contain the actual Copyright notice of the java files they
were produced from not just some generic blurb. This is easy to
implement for the GNU Classpath classes since they always have that
statement as the first comment block at the top of the file. In general
it will not be that easy to extract this info from the java source file.

> Again, comments welcome.
> 
> Without a working AWT I don't believe we can release a 1.0+ version of
> Classpath.  So the version numbers should probably continue to
> simply increment until there is a working AWT.

Seems fine to me. Thanks for taking the lead on this one,

Cheers,

Mark