- From: <bugzilla@wiggum.w3.org>
- Date: Mon, 17 Dec 2007 18:37:03 +0000
- To: public-qt-comments@w3.org
- CC:
http://www.w3.org/Bugs/Public/show_bug.cgi?id=5315 ------- Comment #1 from chamberl@almaden.ibm.com 2007-12-17 18:37 ------- Michael, I am not enthusiastic about allowing all our updating expressions to operate on multiple target nodes. The semantics of such a feature would be quite complex. In general, the target expression of an update may return a mixed sequence of different kinds of nodes (elements, attributes, etc.) The semantics of updates for the various types of nodes are different (for example, you can insert into an element but not into an attribute; you can replace an attribute only with another attribute; renaming an attribute may need special rules for namespaces as you observed in Bug 5316; etc.) Specifying the semantics of updates on sequences of nodes would be a nightmare. It would also contradict the XQuery philosophy (frequently espoused by you, Michael) of simple, composable expressions (e.g., use a FLWOR to update multiple nodes.) You might reasonably argue that, for consistency, we should not allow a delete expression to operate on multiple nodes, even though the semantics of multiple deletions are harmless and straightforward. I would be inclined to retain multiple deletions since they have been in the spec for a long time and are well understood. The work of removing them from the spec and the test suite at this point seems to have a pretty low return on investment. But if consistency is our highest priority, I would much rather remove multiple deletions than try to specify the semantics of multiple-everything-else. --Don Chamberlin (speaking for myself)
Received on Monday, 17 December 2007 18:37:09 UTC