On 22/05/11 12:22, Nikola Smolenski wrote:
On Sat, 2011-05-21 at 12:08 +0100, Neil Harris wrote:
This means that, on modern browsers with both @font-face and UTF-8 support, we should now be in a position to support all of these scripts, even if they are not supported by the user's installed fonts, by adding @font-face declarations for Free Software fonts supporting these scripts, and hosting these fonts on WMF servers.
Just to remind that one more use case is for all Wikipedias to use additional @font-face declarations on pages that contain uncommon characters.
That would be fantastically cool, albeit needing yet another pass over the output text in order to check its characters against the various character sets supported by different fonts.
However, since as far as I know all or almost all of the usable public code points are in Unicode planes 0 and 1, all of this could be achieved by building a single 8k bitmap, and then mashing this together with the corresponding bitmaps for each font. In any case, most articles in most Wikipedia editions would already be dealt with by the default fonts on most platforms, so the overhead would be quite small. We could potentially use user-agent sniffing to detect common operating systems, so we could make a good guess at the default set of fonts already installed.
Also: the Gnu Unifont, although it is a fairly low-resolution bitmap font, is an excellent fallback font for most scripts that don't have elaborate combining rules or font shaping; just serving WOFFs derived from this as @font-face fonts by default for some of the wikis with less-common scripts would be very useful indeed.
-- Neil