Re: [general] testing for 6.0M1?

[Mark Hindess <ma...@googlemail.com>]

In message <4A...@googlemail.com>, Oliver Deakin writes:
>
> Mark Hindess wrote:
> > I've merged the latest commits for 5.0M11 across to the java6 branch
> > at r807351.  Obviously this week we should focus on testing 5.0M11
> > (r807222) so we can hopefully start a vote on Friday but feel free to
> > start testing this revision of the java6 branch so that we can consider
> > whether we could make a 6.0M1 release a week after the 5.0M11 release.
> >
> > I see quite a few test failures but that doesn't mean we shouldn't
> > release.  We have to start somewhere and we can try to reduce the
> > failures for subsequent releases.
> >
> > What do others think?
> >   
> 
> Sounds good to me. I just ran the tests on Windows x86 and I see 81 
> failures and 43 errors. The vast majority are JDWP tests (known and I 
> have raised JIRAs for them - I don't think these should hold up 6.0M1) 
> and TreeMap tests (71 failures and 18 errors).
> 
> The cause of the TreeMap failures is not obvious (could be differences 
> between Java 5 and 6 spec that the tests are not reflecting?), but since 
> they seem centralised and all the other classlib tests seem in good 
> shape, I would be willing to have a JIRA raised for these with a target 
> of fixing them for 6.0M2 and go ahead with the milestone.
> 
> Does anyone see these same failures/agree?

Yes.  That's pretty similar to what I see on linux/x86_64.  I agree it
shouldn't stop us making a release.

Tim, yes.  I agree we should end the code freeze on classlib/trunk and
jdktools/trunk (but obviously *not* drlvm) when we conclude the 5.0M11
vote.

Regards,
 Mark.