On 5/7/16 08:59, Henri Sivonen wrote: > Have I understood correctly that the TrueType font name is not exposed > by the Web Platform, is not used for identification by any Web Font > mechanism and is, therefore, only relevant to identifying system > fonts?
I believe that's correct.
Since TrueType fonts can have multiple name records, it's possible that the fonts that have legacy Mac name records also have Unicode or Windows name records, in which case us dropping support for (non-Roman, non-Cyrillic, non-CJK) legacy Mac font names wouldn't break stuff.
It's highly likely that such fonts also have Unicode and/or Windows name records; or even if they don't have those (some Mac fonts don't), it's likely that they have legacy MacRoman names in addition to the MacHebrew, MacArabic, or whatever. I don't recall ever seeing a font that -only- had legacy, non-(Roman, Cyrillic, CJK) Mac names and nothing else, although in theory it's possible.
However, just because the fonts have other names that will continue to be supported doesn't necessarily mean nothing will break. Note that a font's Unicode or Windows name need not be the same as its legacy Mac name.
The question is whether there's content out there that is relying on the legacy Mac names to specify its fonts; such content could break. Currently, if a page calls for 'font-family: "רעננה"' we'll find and use the Raanana font on OS X. If we remove support for MacHebrew, that will no longer work; the fact that the font -also- has a MacRoman name and could still be accessed as 'font-family: "Raanana"' doesn't solve anything for such content.
How do we find out if we can remove support for (non-Roman, non-Cyrillic, non-CJK) legacy Mac font names?
So the question is not whether fonts with such names are present on people's systems, but how widely such names are used by content to find the fonts it wants. I can think of a couple of possible ways to try and answer that. One is to crawl the web (or use some big archive, or whatever) looking at all the CSS, legacy <font> tags, etc., to find uses of non-(Roman, Cyrillic, CJK) font names. The other would be to "tag" font names when we decode them in Gecko so that we know what encoding they came from, and then have telemetry to tell us what's actually being used.
A third option is to just do it and see if anyone complains. I notice that 'font-family: "רעננה"' doesn't seem to work in either Chrome or Safari today, for instance, so perhaps it's safe to assume that usage on the web is in the minimal-to-nonexistent range.
JK _______________________________________________ dev-platform mailing list dev-platform@lists.mozilla.org https://lists.mozilla.org/listinfo/dev-platform
