Re: [Devel] Compressed PCF fonts

[Keith Packard]

Around 19 o'clock on Feb 4, David Turner wrote:

> I believe that the XFree86 PCF driver doesn't do that. It always
> decodes data on-demand from the compressed file, be it tables,
> properties or glyph data.

The difference is that the XFree86 font mechanism always loads the whole 
font, and it carefully loads it in file order.  The underlying mechanism 
may well support backwards seeks, but they're not used in practice.

My only interest is in allowing people to use existing directories full of 
fonts without needing to reencode them.  I'm attempting to migrate all 
XFree86 font access to FreeType, and that means reading what people are 
currently using.

> The FreeType PCF driver could mimic this behaviour too (and I
> believe it could also include its own LZ77 decompressor), but
> we're speaking of a complete rewrite here..

I don't think it would require a complete rewrite, just a slight IO 
abstraction with some additional code stuffed into the middle of the
process.

> I don't really see any benefit in parsing BDF font files in real-time,
> even less in decompressing and parsing them in real time.. While
> we're at it, why not use XML-based font formats :o)

We do need a format that can cover Unicode outside of BMP -- PCF is 
limited to 16-bit encodings.  If SFNT can handle this, then we're all set.

Keith Packard        XFree86 Core Team        Compaq Cambridge Research Lab