617499 - Add an e10s test to verify behavior of async messages sent by Firefox while the addon process is making a synchronous call

Status: RESOLVED FIXED

(Add-on SDK Graveyard :: General, defect)

Description

[Atul Varma [:atul]]

bsmedberg said that the addon process doesn't do any kind of message queuing, so an async message sent while Firefox is responding to a synchronous call from the addon process should actually be received and acted upon in the addon process before its synchronous call completes. We should verify this though, both so that we know the situation doesn't change, and so that someone reading the test suite has a better idea of what e10s' behavior is.

Comment 1

[Atul Varma [:atul]]

Attachment: Pointer to pull request

Comment 2

[Myk Melez [:myk] [@mykmelez]]

Comment on attachment 496201 [details]
Pointer to pull request

(In reply to comment #0)
> bsmedberg said that the addon process doesn't do any kind of message queuing,
> so an async message sent while Firefox is responding to a synchronous call from
> the addon process should actually be received and acted upon in the addon
> process before its synchronous call completes.

Is it actually true that it should be received, or is it merely the case that it may be received?  It's unclear to me whether an asynchronous message sent from the chrome process must necessarily be received by the content process before the blocking call from addon to chrome returns, so I'm worried about a race condition in this test.

Nevertheless, it seems to work, so r=myk for now.

Comment 3

[Myk Melez [:myk] [@mykmelez]]

Fix commit:

https://github.com/mozilla/addon-sdk/commit/985418ddd8bfd21a22b57609c628461569d14274

Merge commit:

https://github.com/mozilla/addon-sdk/commit/6822f0d7853200cf21e6430701c103a1dfa4b529