900975 - -moz-osx-font-smoothing does not work properly for elements with display:inline

Status: RESOLVED FIXED

(Core :: Graphics: Text, defect)

Description

[Jonathan Kew [:jfkthame]]

See bug 857142 comment 95 and 96.

Testcase:
data:text/html,<p style="font:16px arial;background:black;color:white">one
  <span style="-moz-osx-font-smoothing:grayscale">two

Compare:
data:text/html,<p style="font:16px arial;background:black;color:white">one
  <span style="-moz-osx-font-smoothing:grayscale;display:inline-block">two

Comment 1

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

Is this from failure to force a text run boundary?

Comment 2

[Jonathan Kew [:jfkthame]]

That seems likely, yes. I expect we need to check this property in BuildTextRunsScanner::ContinueTextRunAcrossFrames.

Comment 3

[Jonathan Kew [:jfkthame]]

Attachment: include font-smoothing setting in the nsFont::BaseEquals comparison

Looks like the way to fix this is to respect the smoothing field in nsFont::BaseEquals.

Comment 4

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

Comment on attachment 785103 [details] [diff] [review]
include font-smoothing setting in the nsFont::BaseEquals comparison

r=dbaron if you also remove the smoothing check from nsFont::Equals.

Comment 5

[Jonathan Kew [:jfkthame]]

Attachment: reftest for -moz-osx-font-smoothing on display:inline element

Comment 6

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

Comment on attachment 785121 [details] [diff] [review]
reftest for -moz-osx-font-smoothing on display:inline element

Hopefully the inline-block vs. inline doesn't lead to any rounding issues.

Also worth checking that it actually fails on 10.6 and 10.7 -- it's possible you might need random-if for one or both of them, or just omission of any annotation.

Comment 7

[Jonathan Kew [:jfkthame]]

(In reply to David Baron [:dbaron] (don't cc:, use needinfo? instead) from comment #6)
> Comment on attachment 785121 [details] [diff] [review]
> reftest for -moz-osx-font-smoothing on display:inline element
> 
> Hopefully the inline-block vs. inline doesn't lead to any rounding issues.

I wondered that, too, but it doesn't seem to be a problem.

> Also worth checking that it actually fails on 10.6 and 10.7 -- it's possible
> you might need random-if for one or both of them, or just omission of any
> annotation.

I pushed it to tryserver to check that the annotations worked as expected: https://tbpl.mozilla.org/?tree=Try&rev=6b3da9214137

(Locally, the tests pass for me on 10.7, but that's because my system has subpixel AA enabled by default, which the 10.6/10.7 test slaves apparently don't. If we get that changed, we'll need to update these annotations.)

Comment 8

[Jonathan Kew [:jfkthame]]

https://hg.mozilla.org/integration/mozilla-inbound/rev/3ca6d03fe179
https://hg.mozilla.org/integration/mozilla-inbound/rev/5197eef777dd

Comment 9

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/mozilla-central/rev/3ca6d03fe179
https://hg.mozilla.org/mozilla-central/rev/5197eef777dd