W3C home > Mailing lists > Public > www-style@w3.org > May 2009

Re: CSS3 Web Fonts issue with ?block on downl oad?

From: Adam Twardoch <list.adam@twardoch.com>
Date: Thu, 07 May 2009 20:17:14 +0200
Message-ID: <4A0325AA.9000501@twardoch.com>
To: David Hyatt <hyatt@apple.com>
CC: Boris Zbarsky <bzbarsky@MIT.EDU>, www-style@w3.org
David Hyatt wrote:
> We skip over loading fonts when doing measurement, so we just move on to
> the next fonts in the list.  This allows a developer to specify a very
> similar fallback font, and so ideally those metrics will end up being
> pretty close to your target font.

This is an unlikely scenario. The whole purpose of web fonts is to allow
web developers to use dramatically different fonts than the
Georgia/Verdana duo: more condensed, inclined, handwriting-based etc.
For small-resolution type, the true typographic variation is not in the
shapes (pixels) but indeed, in the proportions.

> I have yet to see any site that uses @font-face that has caused a visual
> jump of elements in WebKit when the font finally loads.  Then again
> sites that even use the feature are obviously rare.

True. So far, Safari is the only major _released_ browser that supports
remote desktop fonts as web fonts. Internet Explorer supports remote EOT
fonts, obviously, but other browsers that support remote desktop fonts
(Firefox 3.5, Opera 10) are not yet released.

There are currently few quality fonts that can be used as web fonts, but
the number will grow. In fact, just five days ago I became part of an
initiative that aims to release 10 quality text families (over 50 fonts)
as freely-licensable web fonts by the end of this year.

> In practice, you don't see any layout jumping, since the pages that use
> downloadable fonts have tended to be well designed.  This is somewhat
> like web sites putting explicit sizes on images to avoid jumping when
> the images load.  If you design your site right, you'll be pretty robust
> against font geometry changes.

Short: 640K ought to be enough for anybody.

Long: this is completely illusoric and is not supported by
typography-aware designers.

In many cases, it's not possible to provide similarly-proportioned
fallback fonts when the designer has chosen to create the website using
fonts of non-Carterian proporions (i.e. fonts that have proportions
obviously different from the Microsoft fonts designed by Matthew Carter:
Verdana and Georgia). If you use a more condensed font, you can easily
fit three columns onto a web page rather than two (of course, you'd use
a larger point size). But the standard operating systems may not ship
with good approximations of those fonts. Just look at script and
calligraphic fonts as an example.

What you're saying is similar to the idea that web pages should only use
216 colors, because those can approximate all possible colors, or that,
if "proxy" images that use 216 colors were displayed instead of the
actual graphics, people wouldn't mind the FOUC.

I must admit that EOT, which provides a mechanism simiar to that of a
"progressive JPEG", is an attractive alternative to desktop font formats
which would solve some of these problems. In short: with properly-made
EOT, a web page could be perceived to "load much faster" than a the same
page that used a remote desktop font with the same design and glyph



Adam Twardoch
| Language Typography Unicode Fonts OpenType
| twardoch.com | silesian.com | fontlab.net

The illegal we do immediately.
The unconstitutional takes a little longer.
(Henry Kissinger)
Received on Thursday, 7 May 2009 18:21:34 UTC

This archive was generated by hypermail 2.3.1 : Monday, 2 May 2016 14:38:26 UTC