Thunderbird 3.next high-level plan

[Dan Mosedale]

http://www.flickr.com/photos/dmose/4789250396/in/set-72157624482878050/
has a redrawing of a picture we drew last week that we think represents
the basic plan for Miramar (Thunderbird 3.next -- I've shied away from
calling it 3.2 for now, because trunk is currently referring to itself
that way, but we want to base Miramar on 1.9.2).

The intent here is to continue to use trunk as our primary development
substrate, and backport high-value features and fixes that need a
release vehicle to Miramar.

Mark and I talked through a bunch of the details that aren't yet nailed
down, particularly around the faster testing releases and localization,
and he's planning to post something next week digging into what is and
isn't known about those parts so that they get sorted out sooner rather
than later.

In the interests of getting better at practicing some of the agile
pieces that we've been preaching lately, I'm going to talk about our
active feature work in a separate post, as some of it is likely to end
up in the 3.2 release vehicle, and some of it not, and we don't yet know
which.

Dan

_______________________________________________
tb-planning mailing list
tb-pl...@mozilla.org
https://mail.mozilla.org/listinfo/tb-planning

[Ben Bucksch]

On 17.07.2010 03:07, Dan Mosedale wrote:
> The intent here is to continue to use trunk as our primary development
> substrate, and backport high-value features and fixes that need a
> release vehicle to Miramar.

I'm confused. So, all the code (not just some changes) which is
currently in trunk, when would that ship? And when is Miramar expected
to ship?

[Kent James]

On 7/17/2010 1:44 AM, Ben Bucksch wrote:
> So, all the code (not just some changes) which is currently in trunk,
> when would that ship? And when is Miramar expected to ship?

As I understand it, trunk would ship about one month after FF ships with
the same backend. The best estimates of trunk shipping would be first
quarter 2011.

And that is the main issue. The TB team wants to be free to ship some
new features before then, and also keep "trunk" working with Mozilla
2.0 The compromise was to imagine a Tb 3.2 based on Mozilla 1.9.2, but
not try to maintain all of the trunk work on the 3.2 branch. The plan
only works if the drivers are fairly conservative as to what gets
backported to 3.2, as otherwise we end up with the nightmare of trying
to do development on 2 branches simultaneously. It also only works if
most of the new features are implemented through extensions, so that 3.2
really only needs the critical hooks that the extensions need.

It is not a ideal plan, but given the realities it is a good compromise.

[Ben Bucksch]

On 17.07.2010 17:02, Kent James wrote:
> On 7/17/2010 1:44 AM, Ben Bucksch wrote:
>> So, all the code (not just some changes) which is currently in trunk,
>> when would that ship?

> [c-c with Mozilla 2.0 vs. 1.9.2]

Is it technically not feasible to have comm-central/trunk working with
both mozilla 1.9.2 and mozilla-central? As I understand, we had this for
a long time? It's no longer feasible or some other overriding reason?

> As I understand it, trunk would ship about one month after FF ships
> with the same backend. The best estimates of trunk shipping would be
> first quarter 2011.

That is terribly late, though. A lot of the trunk code wouldn't be
shipped for almost a year.

[David Bienvenu]

On 7/17/2010 4:23 PM, Ben Bucksch wrote:
>
> On 17.07.2010 17:02, Kent James wrote:
>> On 7/17/2010 1:44 AM, Ben Bucksch wrote:
>>> So, all the code (not just some changes) which is currently in trunk, when would that ship?
>
>> [c-c with Mozilla 2.0 vs. 1.9.2]
>
> Is it technically not feasible to have comm-central/trunk working with both mozilla 1.9.2 and mozilla-central? As I understand, we had this for a long time? It's no
> longer feasible or some other overriding reason?

Component registration has completely changed on mozilla central trunk, which would make it extremely un-fun to support both mozilla 1.9.2 and mozilla-central trunk with
the comm-central trunk. And I think there is a fair amount of other change going on on the mozilla central trunk (e.g., all gecko interfaces are now unfrozen) that is going
to make it harder and harder.

>
>> As I understand it, trunk would ship about one month after FF ships with the same backend. The best estimates of trunk shipping would be first quarter 2011.
>
> That is terribly late, though. A lot of the trunk code wouldn't be shipped for almost a year.

We've had worse :-) If FF ships sooner, we will try to be in a position to ship TB off the trunk sooner as well.

- David

[Dan Mosedale]

On 7/17/10 5:51 PM, David Bienvenu wrote:
> And I think there is a fair amount of other change going on on the
> mozilla central trunk (e.g., all gecko interfaces are now unfrozen)
> that is going to make it harder and harder.
>>
>>> As I understand it, trunk would ship about one month after FF ships
>>> with the same backend. The best estimates of trunk shipping would be
>>> first quarter 2011.
>>>

My understanding is that we wouldn't be committing to shipping 1-month
after FF, but that that would be the first possible time we _could_ ship
off of trunk. I think it's very hard to us make the call about when the
trunk would ship from where we're standing now. I'm interested in
Mark's thoughts here...

>> That is terribly late, though. A lot of the trunk code wouldn't be
>> shipped for almost a year.
> We've had worse :-) If FF ships sooner, we will try to be in a
> position to ship TB off the trunk sooner as well.

Indeed, it'd be nice if we could ship it sooner, but I have yet to see a
proposal that would thread this needle and balance all the competing
interests here.

Dan

[Mark Banner]

On 19/07/2010 23:18, Dan Mosedale wrote:
> On 7/17/10 5:51 PM, David Bienvenu wrote:
>> And I think there is a fair amount of other change going on on the
>> mozilla central trunk (e.g., all gecko interfaces are now unfrozen)
>> that is going to make it harder and harder.
>>>
>>>> As I understand it, trunk would ship about one month after FF ships
>>>> with the same backend. The best estimates of trunk shipping would
>>>> be first quarter 2011.
>>>>
> My understanding is that we wouldn't be committing to shipping 1-month
> after FF, but that that would be the first possible time we _could_
> ship off of trunk. I think it's very hard to us make the call about
> when the trunk would ship from where we're standing now. I'm
> interested in Mark's thoughts here...

That's basically correct. A lot of it will depend on what shape trunk is
in, for example there are lots of bugs around in current trunk
nightlies, and I'm expecting a few more.

Obviously as FF 4.0 gets closer to ship, this will settle down a bit.

Also, this will depend on what features/back-end fixes we've landed,
i.e. do we have enough to make it worth doing a ship that soon? It is,
of course, entirely possible that we will have, but we'll have to wait
and see.

Mark.