upcoming electrolysis plans

Dan Mosedale dmose at mozillamessaging.com
Wed Jul 21 20:18:46 UTC 2010


  There had been some concern that upcoming electrolysis changes 
happening in Gecko-land might effect Thunderbird in a major way sooner 
rather than later. I spent some time chatting with bsmedberg, and it 
appears that there is nothing too imminent that we need to be concerned 
about.

With permission, I've included the IRC log of that chat with this message.
I've sent email to Benjamin as he requested, so hopefully we'll see some 
updated docs about this before too long. If people are interested in 
more detail (eg having a call) after those docs come out, please let me 
know...

Dan

[12:35pm] dmose: over in tb-land, we're trying to get a better 
understanding of what the core electrolysis work is going to mean for 
thunderbird
[12:35pm] dmose: and in particular what sorts of changes we're going to 
need in order to keep working
[12:35pm] dmose: as gecko evolves
[12:35pm] dmose: are there any docs or web pages that would give us 
insight here?
[12:35pm] bsmedberg: In terms of the main *content processes* effort, I 
wouldn't worry about it yet.
[12:36pm] bsmedberg: Dolske doesn't even know how it will affect Firefox.
[12:36pm] bsmedberg: Multi-process jetpack is coming soon.
[12:36pm] bsmedberg: Or at least the ability to have jetpacks run in a 
separate process.
[12:36pm] bsmedberg: And we are going to be requiring libxul, or at 
least a static libxul-like linking configuration.
[12:37pm] dmose: yeah, i think standard8 is hoping to get something 
libxul-like in the next several weeks
[12:38pm] dmose: my second-hand understanding is that mobile is already 
refactoring lots of its front-end stuff, which i assume (correctly?), is 
related to the content-process work.
[12:40pm] bsmedberg: yes
[12:40pm] bsmedberg: Fennec is shipping a content process in Fennec 2 
(October, mozilla 2.0 timeframe)
[12:40pm] bsmedberg: Firefox isn't, not until Firefox 5 next year.
[12:40pm] dmose: ah, ok
[12:41pm] dmose: and are there current plans for any content-process 
related changes that we would be _required_ to deal with?
[12:42pm] bsmedberg: Probably not in the 6-12 month timeframe, no.
[12:43pm] dmose: meaning that such changes could come as early as 6 
months from now, but perhaps as late as 12 months?
[12:43pm] bsmedberg: Well, we can't even think about removing support 
for in-process browsers.
[12:43pm] bsmedberg: Until Firefox ships with content-process browsers.
[12:44pm] bsmedberg: After that, there are engineering decisions about 
whether supporting in-process browsers is holding us back somehow.
[12:44pm] bsmedberg: And I can't predict that far in the future.
[12:45pm] dmose: so if i'm understanding correctly, it's entirely 
possible that supporting in-process browsers could last indefinitely?
[12:45pm] bsmedberg: Yes.
[12:46pm] bsmedberg: There are various things which will probably move 
to other threads or processes, like compositing.
[12:46pm] bsmedberg: But that shouldn't affect you, as long as you're 
building with --enable-ipc
[12:46pm] dmose: good to know
[12:46pm] bsmedberg: That will be required sooner, rather than later, 
which is where the libxul requirement comes from.
[12:46pm] dmose: ok, makes sense
[12:47pm] dmose: the high-level thing here is that i'm trying to figure 
out whether it makes sense to have a phone call sometime soon with you 
and some of the main mail hackers
[12:47pm] dmose: and my impression from this conversation is that it's 
not urgent, which i had been concerned that it might be
[12:47pm] bsmedberg: Yeah, it's not urgent.
[12:47pm] dmose: that said, i suspect it could still be valuable
[12:47pm] dmose: is that something you'd be up for sometime in the 
upcoming weeks?
[12:48pm] bsmedberg: I feel like I want to, and I feel like I don't.
[12:48pm] • dmose chuckles
[12:48pm] dmose: can you unpack that a bit?
[12:48pm] bsmedberg: If there are specific concerns, I think an email 
exchange would work better.
[12:49pm] bsmedberg: If people just want to get a 'feeling' for what's 
going on, I guess a phone call might work better.
[12:49pm] dmose: i think it's more the latter
[12:49pm] bsmedberg: I think it might be a lot of me going "I don't 
know" over and over again
[12:49pm] dmose: heh
[12:49pm] dmose: well, in that case, it would be a short call!
[12:49pm] dmose: alternately, are there docs describing the basic 
current plans that people could look at?
[12:50pm] bsmedberg: Hrm, there are some very out-of-date ones, and 
crumbs that we've been dropping in the platform meeting.
[12:50pm] bsmedberg: Send me an email to remind me, and I can update the 
e10s planning pages to make them fairly current and accurate.
[12:51pm] dmose: ok, will do
[12:51pm] dmose: in the mean time, are you ok with me posting this IRC 
log somewhere public and mailing it to tb-planning?
[12:51pm] dmose: perhaps lightly edited
[12:52pm] bsmedberg: yes
[12:52pm] dmose: great; thanks much!



More information about the tb-planning mailing list