[OpenLayers-Dev] Fwd: 2.7 rc1 -- Friday Morning

[Erik Uzureau]

Dear Dev,

In case it slipped your mind... we have been planning to do the first
RC for the 2.7 release of OpenLayers at the end of this week.

As you all know, the plan for the last two months has been to do the
first RC at the end of August, but we have pushed it back already one
week. Pushing it back any further will not give us the necessary time
to do a series of proper RC cycles and the final release in time for
the FOSS4G2008 Conference... which has been our goal for this release
since the beginning.

Right now, if we look at the tickets page:
http://trac.openlayers.org/wiki/Release/2.7

...we see that there are only 13 tickets left on the plate, 7 of which
are up for review. Of the remaining 6, 3 are related to the
vector-behavior work.

Since the beginning when I proposed the new style for this release
(which was unanimously supported, mind you), the plan has been to
stick to a date at the expense of finishing every last ticket/feature.

The issue we have is that so much work has been done on the
vector-behavior that it seems a shame to leave it out of the release.
At the same time, I have been told that the code, in its current
state, is not really ready to go into trunk.


Which leaves us with two choices:

1) Wait until vector-behavior is done to do 2.7

Doing this not only means that we will not have a 2.7 release in time
for the conference, but, more importantly, negates the entire
experiment we've tried of making a date-based (instead of
feature-based) release. Besides the obvious blow to my ego as the
official wet-noodler, I think that for all of us this is sort of a sad
admission of defeat.


2) Push vector-behavior into the 2.8 release.

Allow for the proper completion of the work and its adaptation into
trunk, and then do a quick, followup 2.8 release. Reasonable target
date could be as early as late October, depending on how quickly the
work can be finished. I will personally volunteer to do the work of
release manager again -- in consolation to those who will be
disappointed/disadvantaged by a 2.7 without vector-behavior. We can do
the 2.8 release quickly and painlessly and I will take responsibility to make
sure it does.



Not being particularly satisfied with either of those options....

Chris Schmidt and I had a phone call this afternoon and he came up
with a good idea for a sort of happy-compromise third option:

3) Do the 2.7 rc1 release on Friday, as planned. Give the folks
working on the vector-behavior a special exception permit -- just this
once! -- to work an extra week (until the rc2, which will be next
Friday) on their patches and bring them up into
rc2. The official OpenLayers policy is that we do not any non-bug
fixes (ie features) after the first RC. However, justification for
this special exception is that since the vector-behavior work is
completely separate (no interaction) from the main code-base.... the
risk of these patches causing any regressions is essentially nil.
Furthermore, since it is a brand new feature, there are no
applications out there that will be testing it for bugs... so again,
it is probably pretty safe. Obviously, the main reason for this
special exception is social and not technical: we don't want to
disappoint any of the developpers who have worked hard on this patch,
but at the same time we do have to stick to our word and follow
through on the rules we set out months ago.



So, at this point, my plan is to suivre option (3), meaning going
ahead with the RC this week. Tomorrow at about noon, I will start
drawing up the release notes and the rc1 announcement. Friday morning,
first thing I do when I wake up will be to put finishing touches on
the notes, make the 2.7 branch, and send the announcement out to the
world.

What this means to you: you either get your patches coded and reviewed
and committed by about 9am Friday, or you get yourself over to Mexico
City by tomorrow night and take me out for too much mescal. For the
vector-behaviour people... you have another week to get your code all
sorted and clean and into trunk -- it goes without saying that if it's
not in and ready by rc2... it ain't goin in. :-)


As always, please voice your opinion on this. We are hoping that
option (3) is a reasonable compromise that will hopefully still allow
us to deliver the new release on time. If you don't think it is the
right solution, please speak up. This is, and has always been, a
democratic project. I like to think that I have the group's authority
in drawing the line as I'm suggesting above, but if that is not the
case... let me know.  (It's a dirty job but someone's gotta do it,
etc).


Finally, I think it is worthy of note at this point that whether the
vector-behavior work makes it in to the release or not... we are at
this point sitting on a 2.7 release that represents the closing of
upwards of 130 tickets... including a ton of bug fixes, a handful of
neat new features, many from new contributors to the project... a new
gallery interface (http://gallery.openlayers.org), a new FAQ interface
(http://faq.openlayers.org), and two new Addins
(http://trac.openlayers.org/wiki/Addins/Bookmark and
http://trac.openlayers.org/wiki/Addins/LoadingPanel).

I think it is a great release and will be happy to show it off with
the rest of the crew at the FOSS4G Conference at the end of the month.
It would be premature to start high-fiving at this point, but we have
seen a really great effort these last two weeks -- I personally
appreciate the vote of confidence to everyone who has been pushing to
get this RC done on time... it is a testament to the unity and the
drive of everyone involved in this project. Thanks!

Erik