- From: Karlsson Kent - keka <keka@im.se>
- Date: Wed, 14 Feb 2001 18:34:15 +0100
- To: "'www-i18n-comments@w3.org'" <www-i18n-comments@w3.org>
Further comments relating to LCC 57 on W3C-normalisation: 1. Differentiate between *numeric character escapes* (like &#...; and &#x...; in XML) and *string include escapes* (which may be character escapes, but may be more general; ENTITY _references_ in XML). It may be that the original intent in the document was to let 'character escapes' mean just 'numeric character escapes'; but entity references must be dealt with, whatever each of these things are called. 2. Reccomend requiring that a 'string include escape' is *defined* such that the string it expands to where used do NOT begin with a combining character, nor a numeric character escape for a combining character. 3. Reccomend requiring that a 'string include escape' (their application, where the string is inserted) cannot be followed by a combining character or a numeric character escape for a combining character. That way 'string include escapes' would not cause any NFC-disruption when expanding the 'string include escape'. Note that there may be a combining character before a 'string include escape', and that the string a 'string include escape' is defined to include may end with a combining character without any problem. XML 1.0 does not enforce this in any way; but I think that the next version of XML should do so to: a. Avoid potential NFC-disruption problems when doing W3C-normalisation followed by expansion of entity references, when the definition of the entity is W3C-normalised. b. Make W3C-normalisation possible without it expanding 'string include escapes'; indeed the definition of a 'string include escape' may be unavailable to the W3C-normalisation process, only to be supplied later; further, it is often the (human) author's choise to use a 'string include escape', and those should not be expanded away by a character normalisation process, but only much later. An alternative could be to say that expansion of 'string include escapes' may turn a W3C-normal (or NFC) original into a result that is not normalised. But that would go against the idea of early getting NFC and keeping it. Expanding a 'string include escape' is an 'editing' operation, but one that is normally done long after any other kind of editing of the document.
Received on Wednesday, 14 February 2001 12:38:01 UTC