904836 - more sharing in style code

Status: NEW

(Core :: CSS Parsing and Computation, defect)

Description

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

There are a bunch of things we can do for more sharing in the style code that would help both performance and footprint.  There's a bit of a plan in bug 768816 comment 7; bugs need to be filed for specific tasks.  This bug can serve as a meta-bug to track them.

Comment 1

[Dietrich Ayala (:dietrich)]

David, is there anything concrete happening here, and is there a timeframe within which you expect to be able to work on it?

Vivien, is this truly required for Haida to move forward?

Comment 2

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

I don't really have a timeframe; these are a bunch of decent-sized projects, and they're not currently at the top of the priority list.  But I'd certainly be interested in knowing why they'd be important.

Comment 3

[Vivien Nicolas (:vingtetun) (:21) - (NOT reading bugmails, needinfo? please)]

(In reply to Dietrich Ayala (:dietrich) from comment #1)
> David, is there anything concrete happening here, and is there a timeframe
> within which you expect to be able to work on it?
> 
> Vivien, is this truly required for Haida to move forward?

Hey,

We can move forward without it, but it will be a nice to have.

David, correct me if I misunderstood what you explained me when you were in Paris, but I understood that the platform currently share user agent stylesheets but not authors stylesheets?

Basically, Haida will have multiple windows opened at the same time (1) for the same application (same manifest url). Each window will likely be of the same origin but with a different url and will likely have a set of common stylesheets.

One other possibility for Haida would be that it has only one window opened, but instead of changing the dom directly, the developer will navigate to a new url (2) (still off the same origin).

In both cases my understanding is that sharing more css should helps load time and reduce the memory footprint. It is unclear that we have a memory benefit for (2) since there will always be one current window, but I assume that some css related data are in some ways stored in the bfcache, so it may help too. But in both case it should save some load time as far as I can tell.

David, please don't hesitate to reach me if my explanations are unclear, of it I misunderstood some stuffs.