Markus Kuhn wrote:
> After all, most of the *-iso10646-1 fonts around
> have a repertoire *much* larger than what your pseudo font mechanism can
> possibly provide, because they are larger than the repertoire of all
> other available X11 fonts.
Yes, but there may be some fonts that contain one or more glyphs that
are not yet available in all the *-iso10646-1 fonts. Also, IBM has
Unicode fonts in AIX, but I don't know how their repertoires compare
with the freely available 10646 fonts.
> Instead of attempting some intelligent fully
> automatic font selection, I would actually prefer a very simple and
> efficient mechanism where I can specify a sequence of fonts A, B, C, and
> if a character is not found in A, then the software looks in B, and if
> it is not there either, it will look into C, and if it is not there
> either, you'll get the replacement character displayed.
Yes, this is what the CSS standard specifies. I have mentioned this a
couple of times already. I'll now give you a URL too:
> I just don't like software that tries to be smarter than me -- it
> usually fails and I loose the feeling of being in control of what is
> going on ... ;-)
When we wrote the first implementation of Unicode pseudo-fonts for X,
CSS was still in its infancy. Now that CSS has matured and Mozilla has
committed to support it, we can put more control in the hands of the
authors and consumers. (CSS style sheets can be provided by authors,
consumers and user agents, and there are certain rules for overriding.)
However, CSS also specifies a number of generic font names like serif,
sans-serif and monospace. These are typically put at the end of a list
of fonts, so that they are used as a last resort. They are supposed to
be provided by the CSS implementation, and they can be thought of as yet
another list of actual fonts to try. If I have understood it correctly,
authors and consumers can specify the lists of fonts to try for these
generic fonts (via @font-face), but the implementation (user agent) is
expected to provide the default lists. The spec says:
The user agent makes (or accesses) a database of relevant font-face
descriptors of all the fonts of which the UA is aware.
So, presumably, the user agent (UA) can "make" a database using
techniques similar to the automatic pseudo-font mechanisms that we
Again, the author and consumer can override this, and the UA's automatic
selection should be used as a last resort (i.e. default).
This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:20:43 EDT