- From: Henri Sivonen <hsivonen@iki.fi>
- Date: Tue, 20 Nov 2007 10:58:06 +0200
- To: Mark Davis <mark.davis@icu-project.org>
- Cc: fantasai <fantasai.lists@inkedblade.net>, "Martin Duerst" <duerst@it.aoyama.ac.jp>, "Addison Phillips" <addison@yahoo-inc.com>, www-international@w3.org, www-style@w3.org
On Nov 19, 2007, at 02:21, Mark Davis wrote: > Having some of the characters be case-insensitive and some be case- > sensitive seems the worst of all the choices. I suspect that this > will lead to no end of user confusion; for example, > • duerst = DUERST > • but dürst != DÜRST > and claims of favoritism for English. This is not a problem when the comparison function applies to identifiers defined by the language itself and the language itself uses only Basic Latin for its identifiers. Those identifiers need to be English-based and Basic Latin-only in order for the language to be suitable for international use. FWIW, as currently drafted, HTML5 is ASCII-case-insensitive as well, so the A-Z range is folded to a-z and after that, code-point-for-code-point comparison is used against lower- case canonical names. For user-defined identifiers, I think making them case-sensitive would be the best option, but I'm not sure if that's backwards-compatible enough. In any case, the comparison function for the built-in identifiers and user-defined identifiers can still be different even if both were case-insensitive in some sense. -- Henri Sivonen hsivonen@iki.fi http://hsivonen.iki.fi/
Received on Tuesday, 20 November 2007 08:58:22 UTC