399391 - font-family should not affect treatment of numeric character references

Status: RESOLVED FIXED

(Core :: Graphics, defect)

Description

[Karl Tomlinson (:karlt)]

Numeric character references should be treated as ISO 10646:
http://www.w3.org/TR/html4/charset.html#h-5.3.1.

If a font-family of a non-Unicode font is specified, then the character references are (incorrectly) interpreted as points in the fonts (system-selected?) cmap.

See attachment 282041 [details] from bug 397288 for a testcase.  This should show two y characters and one psi.

(Departure from the standard for the first y to be displayed as psi may be debatable, but even this is inconsistent with the resolution of bug 33127, or the effort to distinguish quirks mode in bug 195038.)

Comment 1

[Stuart Parmenter]

The behavior of handling symbol fonts is correct on Windows.  Other platforms should look and see what they do with fonts without Unicode character maps and match the behavior on Windows.

Comment 3

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

I disagree about this being invalid.  We used to do this only when nsFont::familyNameQuirks was true, which was when the family name came from a <font> tag and we were in quirks mode.

Comment 4

[Masayuki Nakano [:masayuki] (he/him)(JST, +0900)]

Attachment: proposal patch #1

This patch can fix this bug. But Stuart might not allow the changing.

Comment 5

[Stuart Parmenter]

the whole familyNameQuirks is completely busted.  we should remove it entirely from the tree.

Comment 6

[Stuart Parmenter]

Also, if we really want to have this conversation, it should be in bug 33127 which I've reopened.

Comment 7

[Karl Tomlinson (:karlt)]

Fixed by bug 399636.