IcedTea7 2.0 Branched for Release

[Dr Andrew John Hughes]

On 12:08 Fri 30 Sep     , Mark Wielaard wrote:
> Heay,
> 
> This discussions seems a little overheated.

Indeed.  I don't really expect a suggested release date to be met with
a response as if I'd just declared war.  The original note said "I plan to"
not "will be".  Looking at the archives, that's pretty much how we've
always done this and dates have changed as a result.

I've tried to respond as best I can with simple facts, but it's difficult
given the tone of that e-mail and the unfounded accusations of hijacking.

> You might want to split it up a bit.

The only issue I see related to the release is the need for testing time.
The release has now been delayed to coincide with the security updates on
the IcedTea6 branches which should give more than sufficient time.

Something we do have to consider, going forward, is whether we want to delay
releases for extended periods to allow testing on minority architectures.
I don't believe we've ever decided on a minimum testing period.  I usually
prefer "when we're ready", but, unlike with 6, there is now a heavily active
upstream to consider.

> 
> Clearly picking a data for a release should be properly discussed.

And this thread opened up such a discussion.  You don't get very far with
such a discussion unless you at least propose a release date.

> There
> were more considerations than just "Monday seems a nice day for it". I
> like the explanation of taking a security update data and the 7u2
> integration date into account. Would have been nice to have that in the
> original suggestion send out (and no, I don't think it is strange not
> everybody knows about those dates, please always err on the side of
> people having overlooked such details and just provide them again).

True.  I wouldn't have objected so strongly about Matthias not being aware
of the security errata if it wasn't for him accusing of not being aware
of a public holiday (of which I'm still none than wiser as to which country
that's in).

> In
> the future it would certainly be nice to give more people a bit more
> time to prepare, even though someone has to do it, please give long
> enough advance warning when, so everybody can plan towards a date. 

The initial date was a little close, mainly because I was so fed up that this
has been dragging on so long.  We've rectified that now.

As stated above, we need to decide how to compromise with 7 as there are more
pressures than there were with 6 for timely releases.

> Even though clearly in the end someone just has to do it, and clearly Andrew
> does most of the work, please respect that. 

This was my main issue with Matthias' post; it's extremely negative and is not
once grateful for any of the work done by myself and others.  That's why I
explicitly went to the point of thanking people by name.

> But please also do give
> others who might have less dedicated time a bit more heads up.

It was a negotiable release date.  That said, there is a limit to how much the
plans for this project can be determined by someone who is visibly active on only
a biannual basis.  I can't think of any FOSS projects where you'd be met with a
favourable response if you replied to a release announcement with such a rant,
having being absent for months.
 
> 
> Clearly there are some other (technical and procedural) issues, but lets
> try not to mix them all together in one and the same thread. I get the
> impression that if the subject had just been the date of the release
> there would quickly have been agreement. But mixing it with some other
> lingering issues was not a smart thing.

That was the only subject in the original e-mail.

> 
> It certainly might be good to have a discussion about the use of
> forests, having autobuilders test different/more things, what
> commit/review/release policies are and what breakage is (un)acceptable
> in the various components (clearly the autobuilders can help a bit with
> monitoring that if setup properly). Lets start separate threads for
> that.

True.  There has been plenty of time for discussion of these when they were
originally announced.  These issues are months, if not years old, and were
met with silence when originally posted.  Half the time it feels like I'm
talking to myself.  See:

http://mail.openjdk.java.net/pipermail/distro-pkg-dev/2011-March/013215.html

for an example, which is one of the things I believe Matthias is complaining
about.

I don't expect someone to be making changes on a weekly basis, and
maybe the ChangeLog was a bad choice to use as an activity monitor.
However, I would expect someone to at least keep abreast of what is a
fairly low traffic mailing list, and most of the problems seem to stem
from not doing this.

We're hardly going to move from the forest model now after three years of
work has been invested into it.  Likewise, as far as I can see, the current
release model and commit policies are working fine.  

I don't see any point in going on about whether something should have been done in
a particular way or whether someone should have been more active in the past, as
we can't change what's happened already.  To move forward, it would be useful
to see some actual criticism of what the issues are with things as they stand now.

I make no apologies for taking a leadership role.  Someone has to.

> 
> Thanks,
> 
> Mark

-- 
Andrew :)

Free Java Software Engineer
Red Hat, Inc. (http://www.redhat.com)

Support Free Java!
Contribute to GNU Classpath and IcedTea
http://www.gnu.org/software/classpath
http://icedtea.classpath.org
PGP Key: F5862A37 (https://keys.indymedia.org/)
Fingerprint = EA30 D855 D50F 90CD F54D  0698 0713 C3ED F586 2A37