- From: Andrey Mikhalev <amikhal@abisoft.spb.ru>
- Date: Wed, 21 Oct 2009 14:35:29 +0400 (MSD)
- To: Anne van Kesteren <annevk@opera.com>
- cc: CSS WG <www-style@w3.org>
On Tue, 20 Oct 2009, Anne van Kesteren wrote: >> removing them from rules collection making cssom inconsistent/inadequate. >> the same if changes not affect namespace map (subsequent rule insertion >> broken). > > I don't follow. in compound document where info pieces came from different sources, you cannot guarantee success of Sheet.insertRule(X) unless: - namespace prefixes used in X known - namespace prefixes used in Sheet known (can be examined thru api) and/or - namespace prefix conflicts between X and Sheet can be solved since X is string and cannot be decomposed to prefixes w/o parsing, the only way to solve known prefix conflict is direct manipulation with @namespace rules collection in Sheet (add/replace). >> too high price for 'perfect serialization'. i prefer non-serializable impl: >> - changes in @namespace rules mirrored to namespace map in sheet's >> context immediately; >> - such map used only during parse; >> - any subsequent parser invocation use current state of map. > > This seems a lot worse. better than nothing. > > >
Received on Wednesday, 21 October 2009 10:36:04 UTC