329069 - :first-letter should include grapheme clusters

Status: RESOLVED FIXED

(Core :: Layout: Block and Inline, defect)

Description

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

The code to fix this is the same area as that to fix bug 23605 / bug 45091 / bug 328111, but it's not the same bug.

:first-letter should include grapheme clusters when combining characters are used.  Examples to follow.

Comment 1

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

Attachment: examples

Comment 2

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

Note that these examples don't include any that wouldn't be solved by NFC / KNFC normalization, but such examples exist.

Comment 3

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

See http://www.unicode.org/reports/tr29/ for the definition of grapheme cluster; we probably should have a general mechanism for grapheme cluster iteration since it's very useful for editing.  (And such an iterator could be a superset of dealing with surrogates.)

Comment 4

[Simon Montagu :smontagu]

Fixed by new textframe.

Comment 5

[Karl Tomlinson (:karlt)]

The 329069-1.html reftest now passes even on Linux.
http://hg.mozilla.org/mozilla-central/rev/8ed68afab41d

Correcting the styles in 329069-2 makes it pass on Linux (only).
http://hg.mozilla.org/mozilla-central/rev/7b49ee09259b

Comment 6

[Stefan]

What about Bug #603710 and its testcase 
https://bug603710.bugzilla.mozilla.org/attachment.cgi?id=482618

Comment 7

[Karl Tomlinson (:karlt)]

Yes, it would be good to check in those testcases as reftests.