Thunderbird and Addons

R Kent James kent at caspia.com
Tue Sep 1 00:49:48 UTC 2015


On 8/24/2015 12:43 PM, Gervase Markham wrote:
> The obvious follow-up questions you will then get are:
>
> * The reason Firefox is deprecating XUL/XPCOM addons is because they
> want to move away from those techologies towards things which are more
> webby. Do you then plan to continue to support XUL and XPCOM, even if
> Firefox moves away from these technologies? If not, what does it mean to
> say that you "continue to support binary addons"?
In the particular case of binary addons, what I mean is that we have not 
disabled binary addons for Thunderbird at this point in time, unlike 
Firefox. Yet we have no expectation of continuing this support beyond 
the point where the core Mozilla code makes this difficult, so addon 
authors who rely on binary addons (with myself as an important audience 
here) are on notice that this may break at any time. Generally for 
binary addons, existing known uses all have authors who are close to the 
project, so this is mainly letting any lurkers know what is already 
pretty well understood by those close to the project.

In the case of XUL addons, the issue is that when Mozilla says "12-18 
months" to move away from XUL, what that means for Thunderbird with our 
slower release cycle is that certainly Thunderbird 45 in March of 2016, 
and almost certainly Thunderbird 52 in late December 2016, will continue 
to support XUL-based addons. The earliest Thunderbird release that may 
not support XUL is Thunderbird 59 in October 2017, roughly two years 
from today. It is too early to say what we will support in two years, as 
the technical plans and schedule for the Firefox XUL replacement are not 
clear.

Concerning XPCOM, its replacement is even further in the future than the 
XUL replacement, so it is not meaningful to talk about a schedule for 
that yet, but it will be at least two years and probably more before 
there is any expectation that XPCOM will be replaced in Thunderbird.

> * Will you additionally support some form of stable addon API, analogous
> to the Firefox one, or not?

Thunderbird needs some addon model. I don't think that we can reasonably 
start talking about what that will be until 3rd quarter 2016 when we see 
what Firefox's XUL replacement is going to look like, and after 
Thunderbird 45 is released.

> If the Thunderbird addons ecosystem does not have the problems with
> mal-addons that the Firefox one does, then the major use case for
> signing goes away. But I'd say that the decision should be driven by a
> threat analysis, not by feedback, which will inevitably be from a small
> subset of users, and probably those least likely to be affected by
> mal-addons, and most likely to be affected by signing.
Within Firefox, experience has show this to be a problem. Within 
Thunderbird, experience has shown that this has not been a problem, as 
confirmed by Jorge. And yet because it has been a problem for Firefox, 
it could be a problem for Thunderbird in the future. So while there is a 
threat, it seems to be a significantly smaller threat than what has been 
endured by Firefox already for many years. That's my "threat analysis".

Given all of the other current uncertainty over addon development, I do 
not believe we are well served by adding this requirement on as a 
short-term burden to our addon developers (and to the Mozilla addon 
team). Unsigned addons will almost certainly be supported in the Gecko 
core code as this is required for developer testing in prerelease 
Firefox, so we should continue to have this choice in the timescale 
while XUL addons continue to be viable.

>> This policy must be modified by the caveat "as long as core Mozilla code
>> can be used to support it".
> Which, in the case of XUL/XPCOM, is a major caveat. It would not be good
> for Thunderbird to announce that it plans to continue supporting such
> addons, but in practice it ends up not supporting them on the same
> timescale that Firefox stops supporting them.

The message that I want to put out is how the Firefox announcement 
affects Thunderbird given what we do and do not know. I don't have an 
intention of announcing that we will be supporting XUL addons for the 
foreseeable future, regardless of what Firefox does. We should say what 
we can say rather than remain silent, as people are asking about our 
plans. But my main point is that Thunderbird will likely lag Firefox, 
and try to put some reasonable guesses on how much.

:rkent



More information about the tb-planning mailing list