- From: Michel SUIGNARD <Michel@suignard.com>
- Date: Tue, 2 Jun 2009 10:31:20 -0700
- To: "Michael Day" <mikeday@yeslogic.com>
- Cc: "John Daggett" <jdaggett@mozilla.com>, <www-style@w3.org>
> From Michael Day: > > This is a good improvement I think but maybe it would be better to just > > add a string value to the possible values of <urange>. > > That's a good idea. The definitive source for Unicode block names would > be the Blocks.txt file, eg. for Unicode 5.1: > > http://unicode.org/Public/UNIDATA/Blocks.txt I am not a big fan of the Unicode blocks as a way to define font coverage capability, although I was part of the design team that created them for the True Type tables. Although blocks started out reasonably clean in the early ages of Unicode, they are now created in a way to optimize filling of the Unicode planes (especially BMP: Plane 0) w/o much consideration for identification efficiency, the Latin blocks are probably the worst offenders in that matter with now 6 blocks containing Latin characters. Then you have some characters that are shared among blocks that are essential for others one (like between Arabic and Syriac, and across Hindi blocks). Finally the list of blocks is anything but stable as it grows all the time, including for existing scripts. I think that either using hex ranges (as originally designed) or language script values (such as Latin, Greek, Cyrillic, Arabic, etc... as defined by Unicode UAX#24) is a better approach. And even with that, it requires some skills to create a font content that adequately cover writing systems, because of the shared characters (typically classified as 'Common' or 'Inherited' in term of script values). Michel
Received on Tuesday, 2 June 2009 17:31:55 UTC