W3C home > Mailing lists > Public > www-style@w3.org > March 2010

Re: [css3-fonts] opentype font feature support

From: Mikko Rantalainen <mikko.rantalainen@peda.net>
Date: Fri, 05 Mar 2010 10:52:15 +0200
Message-ID: <4B90C63F.4070804@peda.net>
To: www-style list <www-style@w3.org>
Thomas Phinney wrote:
> On Thu, Mar 4, 2010 at 5:08 PM, Jonathan Kew <jonathan@jfkew.plus.com> wrote:
>> So, I think the essentials would be
>>    font-kerning: none | normal
>> or
>>    font-kerning: no-kern | kern
> I think one problem is that what is "normal" is ambiguous here,
> especially in a universe where most user agents don't know how to kern
> at all (yet).
> "Auto" to me suggests the UA is free to do what it thinks best. This
> might mean kerning at larger sizes only, or only kerning at text sizes
> when the size of the kern exceeds some threshold, or....
> There is also the idea of "auto" as in "override the kerning in the
> font (if there is any) with your own best guess as to correct
> kerning"; this is an option in InDesign, but one doesn't see it in
> much of anything else, and it is reasonably processing-intensive.

How about

	font-kerning: never | auto | always

- never obviously means no kerning,
- always tells user agent to do whatever it can to enable kerning
(including possibly analysis and kerning out of thin air)
- auto lets the user agent to decide (sane default would be to use
kerning for fonts that include the required data and the kerning does
not eat too much CPU - I think that processing requirements should be
the limit more than a font size threshold or something else. auto could
also include kerning through analysis if such analysis is adequately
high quality and processing resources are abundant or the user agent is
e.g. batch processing for highest quality possible)


I believe that there's no need to force "guess the correct hinting". If
some author really wants such feature he can strip the kerning
information out of the downloadable font and then use font-kerning:
always. I have no idea why anybody would want to do that, though. The
result would obviously be totally user agent dependent because there is
not a single correct way to do automatic kerning analysis. For an author
looking for a good result, a sensible way would be to run suitable
analysis for such font file once and then embed the computed information
into the downloadable font.


Received on Friday, 5 March 2010 08:52:54 UTC

This archive was generated by hypermail 2.4.0 : Friday, 25 March 2022 10:07:43 UTC