On 10/26/2015 10:33 PM, Mitchell Baker wrote:
I'm in the middle of a hectic business trip, so will write more while
i"m on the plane home tonight.
The first thing to figure out, as Doug said, is how much support the
project should provide Thunderbird. 5 minutes might be an obvious
yes, as noted earlier. At some level the answer is an obvious "no."
My clear recollection is that Brendan was of the view we had reached
that point a while back, and that Thunderbird should be more separate
from firefox-centric systems, not less. And that he was vehement about
not merging the two repos.
In an ideal world I agree that separating the two so that Thunderbird
team is not spending so much time dealing with Firefox systems is much
better for thunderbird. I also suspect that this will become more
true over time, not less. And I agree that Firefox and our new
products will benefit from having a clean separation, and that we need
all the forward momentum we can get if we are going to impact the
state of the web in the future.
I'm not sure we're able to do this, so need to do some due diligence
to figure out what the ideal world today is. I personally still live
in Thunderbird, so am interested both for the sake of the thunderbird
the open source project, and for my own personal daily experience.
I think there is a general consensus among the Thunderbird developers
that the ideal end goal of Thunderbird would be a heavily JS/HTML-based
system that could be largely independent of Firefox and mozilla-central.
But I should point out that there is a lot of inherent vagueness in that
statement, and part of that vagueness is because it assumes a vision of
a platform that Mozilla wants to provide that doesn't really exist at
the moment.
Even if we agree that's the ideal end state, the problem is that it is
quite a difference from the current reality, and the path to get there
is measured in years. A lot of our current codebase is in C++, and it
requires significant outlay of effort to replace those bits in JS (I say
as one of the people making the herculean effort to actually port
crufty, incomprehensible, poorly-written C++ to easy-to-understand JS
paradigms). The current Gaia Email project has had years of effort to
work on doing a similar task, and the fact that there is still lots of
functionality missing in their work compared to what Thunderbird can do
today is a sign of how hard it would be to get there, even if everyone
worked their hardest and there were no problems caused by
mozilla-central-induced bustage. Unfortunately, the sad reality is that
the antics of Firefox development in recent months and years requires us
to bind ourselves even tighter to Firefox in the short term (by which I
mean the next several years) [1], and I rather suspect that there is no
will (or maybe even ability) on the part of either mozilla-central or
comm-central developers to make the changes required to reverse this
process.
One grave concern I have about this thread and discussion is that people
are being put in a position to decide the life or death of Thunderbird
and are making a decision to keep Thunderbird alive which would in
reality strangle it to death based solely on misinformation and
misapprehensions about the current state of affairs, and I believe that
such a decision would reflect badly on everyone involved. It is
precisely that situation that I am most earnest to avoid, and it is my
sincere belief that merging comm-central into mozilla-central is the
best way forward (and quite possibly the only way forward that would not
lead to the death of Thunderbird, sad as it is for me to admit that).
[1] An example from just this morning is the emasculation of
nsIDOMWindow. It's clear at this point that all of our binary code has
to be linked into libxul and equally clear that mozilla-central
developers are likely to be unwilling to make the changes necessary to
override that decision, and so we need the deep integration into the
buildsystem and the codebase to make that happen.
--
Joshua Cranmer
Thunderbird and DXR developer
Source code archæologist
_______________________________________________
dev-platform mailing list
dev-platform@lists.mozilla.org
https://lists.mozilla.org/listinfo/dev-platform
