668758 - @font-face rules can interfere with generic-font prefs

Status: RESOLVED FIXED

(Core :: Layout: Text and Fonts, defect)

Description

[Jonathan Kew [:jfkthame]]

CSS generics such as 'serif' are mapped to (language-dependent) font family names on the platform via the font preferences such as font.default.serif.x-western.

However, if a @font-face rule happens to use the same family name as was used in the font prefs, the downloaded font will be used instead of the original platform font.

Is this expected behavior? It seems wrong to me; I'd expect the mapping of generics to platform fonts via the preferences to be unaffected by @font-face usage on the page.

Comment 1

[Jonathan Kew [:jfkthame]]

Attachment: example of the unexpected effect of @font-face on prefs fonts

This demonstrates the unexpected interaction on OS X, and should work similarly on Windows AFAICS.

Comment 2

[Jonathan Kew [:jfkthame]]

Attachment: fixed example so it demonstrates the issue on Windows as well

Comment 3

[Jonathan Kew [:jfkthame]]

Attachment: patch, don't look up prefs font families in the user font set

Assuming the current behavior is not actually wanted, here's an attempt to fix it by passing a flag in gfxFontGroup::ForEachFontInternal to tell it whether it should be checking the user font set or not.

Comment 4

[Jonathan Kew [:jfkthame]]

Attachment: reftest

Comment 5

[Jonathan Kew [:jfkthame]]

Attachment: patch, don't look up prefs font families in the user font set

Updated to fix build failure on Linux.

Comment 6

[John Daggett (:jtd)]

Comment on attachment 543525 [details] [diff] [review]
patch, don't look up prefs font families in the user font set

Looks good but we really need to clean up the generic family code at some point, it's somewhat silly that we're parsing "serif" at this level.  dbaron and I talked about this briefly in the past, I think we should be passing in a font list in an array with generics as special tokens, since the style system has already resolved these.  But that's going to take a lot of effort to untangle existing code across platforms.

Comment 7

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

...which is bug 280443 (at least mostly).

Comment 8

[Jonathan Kew [:jfkthame]]

http://hg.mozilla.org/mozilla-central/rev/51b49a7cf2e4 (patch)
http://hg.mozilla.org/mozilla-central/rev/fe33af7f6fa9 (test)

Comment 9

[Jonathan Kew [:jfkthame]]

Argh, this caused a build failure in gfxFT2Fonts on Android.
Backed out:
http://hg.mozilla.org/mozilla-central/rev/854daee4377c
http://hg.mozilla.org/mozilla-central/rev/84bc1d07bacc
http://hg.mozilla.org/mozilla-central/rev/931f06b80727

Comment 10

[Jonathan Kew [:jfkthame]]

Re-landed, with the android bustage fixed:
http://hg.mozilla.org/mozilla-central/rev/812aa60517a7
http://hg.mozilla.org/mozilla-central/rev/64ee37ef692e