Re: Webfont compression

Aryeh Gregor wrote:

> Thomas Phinney wrote:
> > IMO, the main benefit of MTX is not that it is much more compressed
> > than anything else (as you say, it isn't), but that it is the only
> > in-font compression scheme that is compatible with the existing
> > installed base of IE.
> 
> Does IE not support gzip for EOT files, or is there some disadvantage
> to HTTP-level (not in-font) compression that I'm missing?  The disk
> space should be negligible, bandwidth is the only concern I can see.

These days all browsers support gzip compression of *any* web content.
So the real issue is whether MTX or other font-specific compression
techniques offer an advantage over general gzip compression and whether
there's an advantage to using per-table compression as proposed by
Jonathan Kew (i.e. the ZOT format [1]).  Others have noted that breaking
up the glyph table into separately compressed chunks would also have
some advantages (e.g. for commonly used vs. less commonly used glyphs in
CJK fonts).

I was surprised that the issue of compression came up several times
during the Typecon web fonts panel discussion [2] but no one pointed out
that general gzip compression is available for all web content.  Several
folks presented the somewhat distorted view that standard TTF/OTF fonts
have to be served uncompressed while EOT fonts can be compressed.  This
seems to be a persistent misconception.

John Daggett
Mozilla Japan

[1] http://lists.w3.org/Archives/Public/www-font/2009JulSep/0018.html
[2] http://virb.com/typecon/audio/434710

Received on Tuesday, 21 July 2009 20:44:59 UTC