metronome mail for Bazaar 2.0 and 1.18

[Martin Pool]

key dates: 1.18 final 17 August; 2.0b1 or rc1 4 September.

It's all about this page, and coloring it green:

  https://edge.launchpad.net/bzr/+milestone/2.0

Please: keep looking at this page; don't be afraid to target or
escalate bugs you think should block it (though they may be knocked
back); finish things that are in progress; and direct your efforts
towards things on this list.

For users: we'd really appreciate your test results of 2a and 1.18rc1,
either as bugs or even just reports on this list if it's going ok.
For people on Windows or Mac, testing or helping with the installers
will help too.  And from everybody, improving the documentation, or
helping Emma Jane with the web site refresh will make a difference
too.

The 2.0-targeted bugs have progressed pretty well this week, with many
all of the critical (ie "absolutely must do") bugs now fixed, in
progress or merged.  However, there are still some that might block
the release, most notably that we've not yet changed the default
format <https://bugs.edge.launchpad.net/bugs/398668>, and I'd like to
know we've got to the bottom of the upgrade bugs.

I believe Robert's iter changes for commit
<https://bugs.edge.launchpad.net/bugs/347649> is blocked on a
discussion of iter_changes with Aaron, who's offline at the moment.
Let's break that tie this week and let Robert either land it or know
what to change.  Maybe John can do that?

We did 1.18rc1 a week ago.  I don't think this has found any bugs that
need to be fixed before 1.18final so we should roll 1.18final on
Monday.  I'm out of the office Monday so if someone else can roll it
out that would be great; otherwise I will do it Tuesday.

I'd like bzr to dogfood 2a soon
<https://bugs.edge.launchpad.net/bugs/390502>.  Wouter was working on
this but it seems to have stalled blocking on the question of
upgrading stacked branches, which is Ian's
<https://bugs.edge.launchpad.net/bzr/+bug/374735>.  (In the 3.0 user
model let's look at how to do without stacking in the normal case;
it's a pain in the butt.)

It seems like we're having trouble in finding the right balance
between fixing all the bugs and making it as nice as we would really
like, and just getting 2a out.  It's a big improvement for people but
not something they're really going to use widely until we declare it
ready.

We don't generally stall on monthly releases these days, but we do
have some kind of stall in doing a 2.0 release.  It's a big round
number, but it's also just a release, and steady improvement from one
release to the next remains the modus operandi.  So if the upgrade
method is not all we'd like, well, let's fix it, but I won't
necessarily stall the release for that.    People are doing great
rounding-out stuff on testability, documentation, support features,
and I care about that myself, but the 2a repofmt is the guts of the
release.

I've already bumped the in-tree version number to 2.0dev.  We'll do a
new release on the regular four-week schedule ie with an rc or beta on
the 4th September.  I hope this will be the basis for the final.
Whether it is, is up to you too.

We have quite over 20 patches in
<https://code.edge.launchpad.net/bzr/+activereviews> and 8 approved in
<https://code.edge.launchpad.net/bzr/+approvedmerges> - to be honest
not all are urgent but all do represent work that's getting stale.  So
if you can review them, or land them, that's very good.  There's a few
still in bundle buggy, which may be getting a bit neglected.  I think
many are controversial ones for nested trees.

Any comments on this mail are welcome too- too long? missing something
important?

Happy hacking,
-- 
Martin <http://launchpad.net/~mbp/>