Are most of the existing add-ons being sacrificed due to adding support for WebExtension?

Axel Grude axel.grude at gmail.com
Wed Nov 22 14:11:12 UTC 2017


<head> <meta http-equiv="content-type" content="text/html; charset=windows-1252">
> *Subject:*Re: Are most of the existing add-ons being sacrificed due to adding 
> support for WebExtension?
> *From:*Jörg Knobloch <jorgk at jorgk.com>
> *To:*Tb-planning
> *Sent: *Wednesday, 22/11/2017 06:36:50 06:36 GMT ST +0000 [Week 47]
> On 22/11/2017 04:08, Eric Moore wrote:
>> "In general, there have been many interface/IDL changes over the mozilla57/58/59 
>> period and some changes to JS syntax that most add-ons will require some sort of 
>> update to their code. That will distinguish the "maintained" add-ons from the 
>> "unmaintained" add-ons some/most of which will stop working.
>>
>> ....
>>
>> Jörg."
>>
>> Who made the decision to essentially kill off legacy add-ons and when was it made?

no, there was a decison on the Mozilla main code base to stop supporting XPCOM access 
and the XUL technology. This hasn't been made on the mail dev side; we are still 
dealing with the code changes floating across via the shared code base from 
mozilla-central.

AFAIK there is no short term plan to phase out these technologies from Thunderbird 
Addons. Their hosting will be moved to a different domain (not remain on 
addons.mozilla.org), and obviously this will open up a whole slew of other problems 
(who will review source code for security?). As regards code, Thunderbird will have to 
make the decision of decoupling the code (forking) from Mozilla soon or otherwise 
render our legacy Addons unusable.

> That legacy add-ons will be "killed off" is your interpretation. The truth is, that 
> most add-ons will need updates to their code. That's normal, since interfaces to 
> Mozilla-core change over time. Like nsILocalFile was replaced with nsIFile years 
> ago, so no add-on had to break when nsILocalFile was finally removed recently. It is 
> true though that many interfaces have changed recently as Firefox was moving to 
> Quantum (which appears to be quite a success for Mozilla).
>
> Personally I can tell you that I'm using only a handful of add-ons and I've informed 
> authors of changes so all the ones I use still work at TB 57.
It's quite a bit of work keeping up but it is not fundamentally different what we have 
done over thee years anyway. What's a little bit painful are the (necessary) 
innovations in JavaScript which lead to breaking changes that will make Postbox 
compatibility (which has a /very/ old codebase) on an unforked ADdon pretty much 
impossible. However Thunderbird still supports XPCOM and XUL which are so fundamental 
to legacy and the huge customisability of our Addons that removing them makes low 
level stuff impossible (as experienced with my Fx addon QuickPAsswords - I have looked 
into ways of recreating it as WebExtension and so far am very pessimistic as it is 
impossible with the current APIS. We cannot evene open a tab with the stored passwords 
as it is on a privileged URL, there is no API for |nsILoginManager 
<https://developer.mozilla.org/en-US/docs/Mozilla/Tech/XPCOM/Reference/Interface/nsILoginManager> 
|effectively shutting of low level access to anything password / credential related)

So as long as Thunderbird allows XPCOM our Addons are safe and can be kept compatible 
with a reasonable amount of effort. Of course it helps to have some people on the Tb 
team who reach out and help - Jörg is one of the most importat people as he is 
developing core code and also cares about Addon authors.

> The only alternative to following the interface changes is to fork a version where 
> most add-ons still work. That would be mozilla57. There has been much discussion 
> about a fork. In general, the Council, the Engineering Steering Committee and the 
> developer community all agree that a fork is undesirable, since we have neither 
> knowledge nor manpower to keep up with security patches. So the consensus is to try 
> making it to mozilla59 ESR and then keep going as long as possible without a fork. 
I agree that this buys us some time (even though it is slightly awkward for use Addon 
authors) but the move to a fork is ultimately unavoidable. If parts of the XPCOM 
interfaces are rewritten or dropped now at a more dramatic rate we are going to get a 
hard time finding alternative ways of coding workarounds. ALso what we will see is a 
lot of dropping of backwards compatibility, again something that becomes unavoidable 
as the code progresses.

Axel

*Axel Grude <mailto:axel.grude at gmail.com>*
Music Production and Composition
Thunderbird Add-ons Developer (QuickFolders 
<https://addons.mozilla.org/thunderbird/addon/quickfolders-tabbed-folders/>, 
quickFilters <https://addons.mozilla.org/thunderbird/addon/quickfilters/>, 
QuickPasswords <https://addons.mozilla.org/firefox/addon/quickpasswords/>, Zombie Keys 
<https://addons.mozilla.org/thunderbird/addon/zombie-keys/>, SmartTemplate4 
<https://addons.mozilla.org/thunderbird/addon/smarttemplate4/>)
Visit my YouTube Channel <https://www.youtube.com/c/thunderbirddaily> for email 
productivity tips Get Thunderbird!
>>
>> Who was aware of that decision?
>>
>> Why wasn't this treated as a strategic decision that needed feedback/discussion 
>> like was done with getting a new home? 
>
> Hi Eric,
>
> what you write sounds like someone on the Thunderbird project has made decisions 
> which negatively affect TB. That is not the case.
>
> That legacy add-ons will be "killed off" is your interpretation. The truth is, that 
> most add-ons will need updates to their code. That's normal, since interfaces to 
> Mozilla-core change over time. Like nsILocalFile was replaced with nsIFile years 
> ago, so no add-on had to break when nsILocalFile was finally removed recently. It is 
> true though that many interfaces have changed recently as Firefox was moving to 
> Quantum (which appears to be quite a success for Mozilla).
>
> Personally I can tell you that I'm using only a handful of add-ons and I've informed 
> authors of changes so all the ones I use still work at TB 57.
>
> No one has made any decision, the facts are imposed on us by changes in 
> Mozilla-core. All people actively involved with the development of TB were aware of 
> the never-ending bustage coming in from Mozilla-core.
>
> The only alternative to following the interface changes is to fork a version where 
> most add-ons still work. That would be mozilla57. There has been much discussion 
> about a fork. In general, the Council, the Engineering Steering Committee and the 
> developer community all agree that a fork is undesirable, since we have neither 
> knowledge nor manpower to keep up with security patches. So the consensus is to try 
> making it to mozilla59 ESR and then keep going as long as possible without a fork.
>
> I hope that answers your query.
>
> Jörg.
>
> _______________________________________________
> tb-planning mailing list
> tb-planning at mozilla.org
> https://mail.mozilla.org/listinfo/tb-planning


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.mozilla.org/pipermail/tb-planning/attachments/20171122/ba35f173/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: thunderbird_blog2.png
Type: image/png
Size: 846 bytes
Desc: not available
URL: <http://mail.mozilla.org/pipermail/tb-planning/attachments/20171122/ba35f173/attachment.png>


More information about the tb-planning mailing list