939688 - Make sure the local transport actually freezes the packets

Status: RESOLVED WONTFIX

(DevTools :: Framework, defect, P5)

Description

[Panos Astithas (he/him) [:past] (please ni?)]

Attachment: WIP v1

Jim had told me a long time ago that LDT_deepFreeze was broken and doesn't actually freeze the object's properties. It turns out that the broken behavior is something some actors started to rely on, so fixing it is more work than I expected. I am attaching my current WIP which passes all xpcshell tests, but fails many chrome mochitests. Haven't tried browser mochitests yet.

Comment 1

[Panos Astithas (he/him) [:past] (please ni?)]

Attachment: WIP v2

Fixed most inspector tests, more to come still.

Comment 2

[Panos Astithas (he/him) [:past] (please ni?)]

Attachment: WIP v3

More to come.

Comment 3

[Panos Astithas (he/him) [:past] (please ni?)]

Attachment: WIP v4

More test fixes, but I've stumbled across bug 814892, which blocks further progress, since the App Manager uses a proxy in its DeviceStore. There are a few more weird test timeouts without errors that I haven't investigated yet, but I'm postponing this work until bug 814892 is fixed.

Comment 4

[Eddy Bruel [:ejpbruel]]

Bug 814892 has since been fixed. Panos, I assume you are very busy with your current projects right now, but do you have any plans to return to this bug in the future?

Comment 5

[Panos Astithas (he/him) [:past] (please ni?)]

I plan to revisit this, but it's not high on my TODO list. e10s will make this even less important.

Comment 6

[Panos Astithas (he/him) [:past] (please ni?)]

Attachment: Patch v5

Rebased, tested locally and pushed to try:
https://treeherder.mozilla.org/#/jobs?repo=try&revision=a34e5c834268

Comment 7

[Panos Astithas (he/him) [:past] (please ni?)]

Attachment: Patch v6

Fixed some of the test failures, but I don't really have more time right now for this game of whack-a-mole.

Comment 8

[Jason Laster [:jlast]]

Alex, do you think this is still relevant?

Comment 9

[Alexandre Poirot [:ochameau]]

The deep freeze code added in bug 790202 might still not correctly freeze all nested properties,
but as this only applies to parent process to parent process packet it is quite limited.

I'm not sure it is worth the efforts as most packets are now going through JSWindowActor.

https://searchfox.org/mozilla-central/rev/fa69d8b248e6c1df670aa6b019e30ec37e6672be/devtools/shared/transport/local-transport.js#184-203