|
|
Posted by Alan J. Flavell on 05/06/06 15:05
On Sat, 6 May 2006, Jukka K. Korpela wrote:
> Moreover, no "web-safe" typefaces are needed. Some typefaces are
> available more often than others, but how relevant is that?
One should also mention character repertoire. By naming *some* fonts
(e.g Arial Unicode MS, or Code2000), one can guarantee that if the
font is present then it will have a wide repertoire (although these
particular fonts may be sub-optimal for other reasons). On the other
hand, the commonly-used fonts such as Arial or Times New Roman exist
in a number of different versions with widely differing repertoires,
and some browsers (in particular, the operating system component that
thinks it's a web browser) can be severely hampered in their rendering
of less-common characters if they're directed by the author to use
such a font, rather than being allowed to use the best font which the
reader has.
The particular risk here is that the author might have latest-version
of their chosen font, and might have assured themselves that the page
works to their satisfaction. But the reader might have an older
version of the same font, with a more restricted repertoire, and
lacking some of the characters which the author has used.
http://www.microsoft.com/typography/fonts/default.aspx
There's no way to say in a stylesheet e.g "Use the Arial font, but
only if the version is 2.00(WGL4) or better".
This effect is a particular problem with IE, as I said - at least up
to and including version 6. It's possible to work-around the problem
by referencing IE-specific downloadable fonts (WEFT), while leaving
other browsers to do the better job that they already do (picking the
missing characters out of some other available font) - but although I
have documented this approach, it's not one that I really recommend
for widespread use.
Rumour has it that IE7beta has addressed this shortcoming (I haven't
tried it myself), but AIUI, XP SP2 will continue to be a minimum
requirement for running IE7, so there will be Windows platforms in use
for quite some years yet which won't be able to run IE7.
> It _would_ be possible to make useful heuristic checks on typeface
> names, though. Some typefaces, most notably Symbol but also
> Wingdings fonts, are almost certainly used in an attempt to hack
> browsers into rendering characters incorrectly, so it would be quite
> adequate to issue a warning.
Good point. Also the mischievously-named "Webdings" font, which might
be good for a lot of things but is utterly INappropriate for the web.
> Some typefaces, especially Code2000, look very poor without font
> smoothing.
Thanks for that comment. Indeed I just tried enabling font smoothing
on this Win2K system - where it isn't the default, and the Code2000
characters went from looking black but rather crude, to looking quite
presentable but rather grey and somewhat fuzzy at a normal viewing
size. AIUI, Code2000 is intended to provide an enhanced character
repertoire with a modest-sized font file, so that's inevitably going
to be a compromise.
[...some good points snipped without further comment...]
> ... or helvitca for Helvetica. :-) Of course, it would not matter
> much here, since browsers that don't have Arial would here ignore
> helvitca as unrecognize
However, this does bring us to the question of Helvetica versus Arial,
and - at least in the past - I've encountered three different
scenarios reported by users who have both fonts. I'm no expert in
font management, but I pass these on with usual disclaimers.
1. users who have a real Helvetica font, which is better than their
Arial font
2. users who have a hacked-up Helvetica font because some (non-web)
documents insisted on one - but their Arial font is better
3. users who have Helvetica and Arial recognised as aliases for the
same font
So, putting Helvetica in CSS as first choice would be preferable for
users of type 1 - harmless for most users (who don't have Helvetica
and would just skip to the next choice) - and harmful to users of type
2.
Whereas, putting Helvetica after Arial would be preferable for users
of type 2 - harmless for most users - and harmful to users of type 1.
Users of type 3 don't care, of course.
A relatively small proportion of users are affected by this issue, but
they are, IME, typically users who feel that they have made a
discerning choice (e.g to buy a non-mainstream plaform and/or
browser), and tend to be quite annoyed at getting their discerning
choice trampled on by some undiscerning author.
Do you know - it's suddenly occurred to me that maybe this ought to be
like content-type negotiation, with authors able to specify a
preference factor for each of their listed font preferences, and
readers could specify a quality factor for the fonts which they
possess. The browser would then compute the best compromise between
them. Just a thought, anyway.
Navigation:
[Reply to this message]
|