1623791 - Remove segment gridding optimization

Status: RESOLVED FIXED

(Core :: Graphics: WebRender, task)

Comment 1

[Glenn Watson [:gw]]

Reference to why image tiles currently use segments - https://bugzilla.mozilla.org/show_bug.cgi?id=1526235

Comment 2

[Glenn Watson [:gw]]

Notes to self on things we want to consider:

• 9 patch segment / instance format.
• Unifying glyph / primitive instance format.
• How to handle texture UV sub-rects (and what these are currently used for).
• Not using segment infrastructure for image tiling rendering.
• A simpler / smaller instance format (less indirection?).
• Making use of (unifying) this functionality for image borders, which are already 9-patches.

Comment 3

[Glenn Watson [:gw]]

• A more advanced batch instance container that can handle different instance sizes per shader - this would allow, for instance, much smaller instances for most primitives, and a larger instance for YUV primitives.

Comment 4

[Glenn Watson [:gw]]

• Primitive headers indirection is not ideal - we could remove this once segmentation is simplified to be 9-patches only. However, we need a solution for glyph rendering, since that has a similar issue (many sub-rects per overall text run primitive).

Comment 5

[Glenn Watson [:gw]]

Attachment: Bug 1623791 - Remove segment gridding optimization.

This is no longer as important, with picture caching. Removing it
will simplify the planned changes to switch to a simpler segment
model based on nine-patch rectangles during scene building.

Comment 6

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/5d5fee77d845
Remove segment gridding optimization. r=nical

Comment 7

[Stefan Hindli [:stefan_hindli]]

https://hg.mozilla.org/mozilla-central/rev/5d5fee77d845

Comment 8

[BugBot [:suhaib / :marco/ :calixte]]

The leave-open keyword is there and there is no activity for 6 months.
:gw, maybe it's time to close this bug?

Comment 9

[Glenn Watson [:gw]]

This is still relevant, but low down the priority list for now.

Comment 10

[Mathew Hodson]

Since a patch was landed and there are regressions tracked against that change, maybe this bug should be closed and a new bug can be opened for followup work. It's confusing when this is listed as the regressing bug, but it's stilled marked as NEW.