Re: Accessing glyphs via symbol char map

[Antoine Leca]

Hi folks,

David Turner wrote:
> 
> Well, according to the TrueType specification, the "symbol" encoding is
> used for "Undefined character set or indexing scheme". Basically, this
> means that the application must know how the glyphs are encoded within
> the fonts, or in this case, that you should add the 0xF000 yourself to
> your application..

I may be wrong, but I seem to remember that 0xF000 is only one of the
possible value (but this is certainly the most used).

>From what I recall, the "correct" way is to look at the OS/2 table, to
take the usFirstCharIndex field, and to map this character as the space
character, so in other words, use  (os2->usFirstCharIndex-0x20) 
as offset. I do not remember where I saw that (probably some place at MS
doc), but this is what Windows 9X appears to do these days.

Of course, it will fail the day when someone will try to access glyphs
below 0x20 in a "symbol" font. However, *IF* Microsoft still do what they
are doing presently (using the OS2 field), it will be useless with Windows
as well, which in pratical term means that the font will be useless anyway.

 
> We could provide this automatic translation to the core library though,
> because many (old) Windows font use this so-called encoding.
> 
> what's the opinion of other developers/users on the mailing list ??

I do not like "automatic" translation very much, because it proves sometimes
difficult to escape them the day you encounter a font that do not fit the
scheme envisioned in the "automatism". So I do like very much "debrayable"
automatism (David, can you provide some translation for my ugly French word?)



Antoine