1533949 - [Fission] Make BrowserChild ready for Fission

Status: RESOLVED FIXED

(Toolkit :: General, enhancement, P3)

Description

[:Felipe Gomes (needinfo for replies!)]

Make the code in BrowserChild.jsm ready to work with out-of-process iframes

Comment 1

[Mike Conley (:mconley) (:⚙️) (PTO: May 24 - June 2nd)]

Attachment: Bug 1533949 - [WIP] Port BrowserChild to JS Window Actors. r?felipe

Comment 2

[Mike Conley (:mconley) (:⚙️) (PTO: May 24 - June 2nd)]

Hey jdai, I have a patch up here that'll port one of our old-school actors to JS Window Actors, but I've realized that that's not going to be a full Fission-compatible port in this case.

BrowserParent / BrowserChild are responsible for determining whether or not a tab can be unloaded, or if a page is blocking that unload via the beforeunload event.

For multiple frames across multiple processes, I think we're going to need a way of broadcasting a message to all of the subframes, and then getting a result from each.

Is that broadcast capability planned?

Comment 3

[:Felipe Gomes (needinfo for replies!)]

hey, the patch is missing the BrowserParent.jsm file!

fwiw, the browsingContext in the parent side has a getChildren() function that can be used to enumerate all children BCs* and then use them to call .getActor().. At least, that was the original plan for broadcasting.. I don't know if some built-in functionality is also planned.

• note that one call to getChildren is not enough.. it requires tree traversal as the items returned could themselves have children.

Comment 4

[John Dai[:jdai]]

Since :Felipe already answered your question, I'm going to cancel the needinfo request. Feel free to open a bug, if you need any functionality. I'm willing to help. :)

Comment 5

[Mike Conley (:mconley) (:⚙️) (PTO: May 24 - June 2nd)]

Attachment: Bug 1533949 - Make BrowserChild functions Fission-compatible, and move to BrowserElementChild. r?NeilDeakin

Comment 6

[Mike Conley (:mconley) (:⚙️) (PTO: May 24 - June 2nd)]

Attachment: Bug 1533949 - Make nsIRemoteTab.hasBeforeUnload check each OOP subframe as well. r?rhunt

Comment 7

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/2d6ba0372205
Make nsIRemoteTab.hasBeforeUnload check each OOP subframe as well. r=rhunt
https://hg.mozilla.org/integration/autoland/rev/574775878141
Make BrowserChild functions Fission-compatible, and move to BrowserElementChild. r=NeilDeakin

Comment 8

[Sebastian Hengst [:aryx] (needinfo on intermittent or backout)]

https://hg.mozilla.org/mozilla-central/rev/2d6ba0372205
https://hg.mozilla.org/mozilla-central/rev/574775878141