[OpenFontLibrary] Fwd: www-font: WebOTF Proposal

[Nicolas Mailhot]

Le Jeu 6 août 2009 22:43, Ben Weiner a écrit :
> Hi there,

Hi,

> With Nicolas Spalinger's comments about font metadata (13:00 BST today)
> fresh in my mind, I want to ask what the open font community thinks
> about this proposal. As I understand it, this proposal is intended to
> *replace* OTF/TTF font linking. Unlike another proposal, it does not
> allow any kind of backward compatibility with any version of 'EOT',
> Microsoft's system for font linking in Internet Explorer from version 4
> upwards.

I am personally not in favour of a web-only font format. Every other web
content uses standard formats and the web is successful because moving
content to and from the web does not have any barrier.

That being said, this particular proposal is even worse than the previous
ones since it makes it easy to create web fonts with site or web-specific
names and ids. The whole FLOSS desktop font stack is built over extensive
logic to substitute fonts whenever appropriate and possible. If you allow
web fonts to use different ids than normal font files, you strip a large
part of the info that makes this substitution effective.

People will claim their web fonts will be perfect and do not need
substitution, but we consider than the user is in control and should be
given the means to refuse a font download if he wants to (to save
bandwidth, for licensing reasons, because the web font is ugly or broken
or stale or not refreshed to take into account new text added on the web
site, etc). Also we've seen all too often web site creators that do not
take our client specificities into account and assume Microsoft or Apple
fonts are available to complete their design. We can workaround it with
smart font management code but its efficiency will be greatly reduced in
they have little data to make decisions from.

For example, when TTF/OTF linking was imlemented in Firefox, Mozilla
developers posted a set of examples. In one of them a well-known FLOSS
font was used instead of Helvetica. But because the css rules linked the
font file directly and never mentioned its real name, browsers would
download it even if another (possibly more recent and better) version was
already installed locally. The writer obviously considered the presence of
this font unlikely, which is true for Apple and Microsoft systems, but
false for Linux systems.

-- 
Nicolas Mailhot