Re: [Freetype] font catalog

[Brian Stell]

Antoine Leca wrote:
> ...
> > Moz first looks for fonts in the document's language group. For
> > glyphs not found in those fonts Moz next looks in fonts in the
> > user's locale's language group. If a glyph is still not found
> > then Mozilla looks in all fonts.
> 
> Good. And where is MS Gothic? Supposedly, in the first two groups
> for Japanese, and this is not a problem, since any offending
> characters (the ones with associated glyph *but* the glyph is empty)
> are *not* Japanese ones (or else the logic is flawed).

For Japanese documents and locales this would work.

Consider non-CJK documents/locales like Latin1. If Chinese 
characters are included via NCR there is often no way to 
dis-ambiguate the Han range. In cases like this MS Gothic could 
well be used on non-Japanese characters producing blank 
characters.

> So you only need to have a correct mapping from Unicode codepoints
> to existing glyphs only when you have "to look in all fonts" (BTW,
> since this is a very uncommon situation, you may probably defer the
> building of the "font database" as long as you do not have to
> render glyphs outside the language groups...) 

The "look in all fonts" definitely does occur. All a document 
needs to do is include a NCR for an unused code point. While we
would all want to say that this is a broken page; if Mozilla
took 90 seconds to search for the glyph (even once) most users 
would assume Mozilla was hung after 30 seconds. Mozilla bug 
reports would come in. No amount of saying it is a broken web 
page would stop the bug reports. People expect the browser to 
work reasonably well even with broken pages.

-- 
Brian Stell
mailto:address@hidden