1690476 - Handle backward compatibility for print/colorscheme simulation changes from Bug 1591120

Status: RESOLVED FIXED

(Core :: Layout, defect, P3)

Description

[Julian Descottes [:jdescottes]]

See discussion on https://phabricator.services.mozilla.com/D103782 and guidelines in https://phabricator.services.mozilla.com/D103782#inline-580995

Comment 1

[Emilio Cobos Álvarez (:emilio)]

I'm happy to do this work btw :)

So, one thing that came up during related work is that it'd be nice if these things (or at least serviceWorkersTestingEnabled) would be set from the parent process only. Do you know how it might affect this?

Comment 2

[Julian Descottes [:jdescottes]]

I'm happy to do this work btw :)

Thanks Emilio! I don't plan to work on it immediately, feel free to take it.
The easiest to test backward compat is to start a Firefox release with --start-debugger-server and then connect to it from about:debugging running in your local build. Then try to debug a tab and check that the print simulation button is hidden.

So, one thing that came up during related work is that it'd be nice if these things (or at least serviceWorkersTestingEnabled)
would be set from the parent process only. Do you know how it might affect this?

If we want to move reconfigure options to the parent process, we have several options.

We could open a similar API on a DevTools actor living in the parent process. For instance, we have target descriptors for our top level targets, which live in the parent process. We can add a reconfigure method on the tab descriptor and this would run in the parent process. (or on another parent process actor, eg root actor).

We will also soon introduce a mechanism of "config flags" that will live next to other global devtools settings (eg today: list of watched resources & targets, list of breakpoints, etc...). Those config flags should be available "everywhere": parent process & content process. After that, it will be much easier to move some logic from the content process to the parent process. (and would not break backward compat).

I'm supposed to introduce the pattern as a prerequisite for Bug 1644397, I hope to get it done this week.

Comment 3

[Emilio Cobos Álvarez (:emilio)]

Alright, should I wait for that bug to look at this?

Comment 4

[Julian Descottes [:jdescottes]]

(In reply to Emilio Cobos Álvarez (:emilio) from comment #3)

Alright, should I wait for that bug to look at this?

Depends on what you mean by "this" :)

If "this" is "setting serviceWorkersTestingEnabled in the parent process" then, yes we should wait on Bug 1690698 (filed yesterday, just focused on adding the config flags mechanism).

But if "this" referred to this bug, I think we should still handle it soon-ish.
For now, if a user tries the print simulation with remote debugging from Nightly to any older channel, it will silently fail, which is not great.
(but nothing is badly broken, the feature is just not working)

Comment 5

[Emilio Cobos Álvarez (:emilio)]

Alrighty.

Comment 6

[Emilio Cobos Álvarez (:emilio)]

Attachment: Bug 1690476 - Deal with backwards compat for simulation features. r=jdescottes

Comment 7

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/ffd5979ffbd6
Deal with backwards compat for simulation features. r=jdescottes

Comment 8

[Bogdan Tara[:bogdan_tara | bogdant]]

https://hg.mozilla.org/mozilla-central/rev/ffd5979ffbd6