1763928 - Bad font smoothing / fuzzy text on offerup messaging pages

Status: RESOLVED WORKSFORME

(Core :: Graphics: WebRender, defect)

Description

[Asif Youssuff]

Attachment: bad offerup.png

Noticed this for a little bit.

Steps to reproduce:

1. Create offerup account: https://offerup.com
2. Message a seller

What happens:

The text looks strange / badly smoothed. This is most obvious in the "chat bubbles", especially in the white text on the green background.

This is a regression.

I captured a singlefile of a message page to make it easier to share (not everyone is going to want to sign up for this service).

Screenshots of good and bad are also attached.

10:34.35 INFO: Narrowed integration regression window from [8e9c1ffd, 41535e3d] (3 builds) to [8e9c1ffd, b7f88e5c] (2 builds) (~1 steps left)
10:34.35 INFO: No more integration revisions, bisection finished.
10:34.35 INFO: Last good revision: 8e9c1ffdeed84e460a9ba7f9e83ca24319a15c87
10:34.35 INFO: First bad revision: b7f88e5c537bb7d64ba9d4cb58b052650d510638
10:34.35 INFO: Pushlog:
https://hg.mozilla.org/integration/autoland/pushloghtml?fromchange=8e9c1ffdeed84e460a9ba7f9e83ca24319a15c87&tochange=b7f88e5c537bb7d64ba9d4cb58b052650d510638

Comment 1

[Asif Youssuff]

Attachment: good offerup.png

Comment 2

[Asif Youssuff]

Attachment: test case bad smoothing offerup

Comment 3

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

Set release status flags based on info from the regressing bug 1749380

Comment 4

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

:gw, since you are the author of the regressor, bug 1749380, could you take a look?
For more information, please visit auto_nag documentation.

Comment 5

[Gregory Pappas [:gregp]]

Attachment: bug1763928_reduced.html

I did my best to reduce this. It's a really fragile bug so I hope it's still reproducible with this file. It seems to preserve the bug on my machine. Does this reduced version reproduce the bug on your machine?

Comment 6

[Asif Youssuff]

(In reply to G from comment #5)

I did my best to reduce this. It's a really fragile bug so I hope it's still reproducible with this file. It seems to preserve the bug on my machine. Does this reduced version reproduce the bug on your machine?

Thanks G, yes - it still reproduces the bug on my machine.

Comment 7

[Glenn Watson [:gw]]

The issue here is that there are two rotateX(-180deg) transforms. The first one affects the off-screen surface that is created due to the border-radius of 4px around the enclosing div. Since that transform results in a z-flip, we don't currently detect it as being in the same 2d coord system that can have snapping applied. The z-flip doesn't actually matter here, since we're in a 2d rendering context only. I need to work out how to correctly detect this case, which is quite subtle due to how complex the current WR z-flattening code is.

Comment 8

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

The severity field is not set for this bug.
:gw, could you have a look please?

For more information, please visit auto_nag documentation.

Comment 10

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

The severity field for this bug is set to S4. However, the following bug duplicate has higher severity:

• Bug 1767579: S3

:gw, could you consider increasing the severity of this bug to S3?

For more information, please visit auto_nag documentation.

Comment 11

[Glenn Watson [:gw]]

I think this is probably resolved on nightly with all the recent snapping work. The test case looks OK to me?