25446 2014-04-24 22:25:34 +0000 [F+O 3.1] ordered collation keys 2014-09-15 09:15:07 +0000 1 1 1 Unclassified XPath / XQuery / XSLT Functions and Operators 3.1 Working drafts PC All CLOSED FIXED P2 normal --- 1 mike mike josh.spiegel public-qt-comments oldest_to_newest 104378 0 mike 2014-04-24 22:25:34 +0000 The new collation-key() function states: This specification does not mandate that collation keys should retain ordering. This is partly because the primary use case is for maps, where only equality comparisons are required, and partly to allow the use of binary data types (which are currently unordered types) for the result. The specification may be revised in a future release to specify that ordering is preserved. However, we have introduced support for ordering in hexBinary and base64Binary, and UCA defines an algorithm for generating binary collation keys that respect ordering. So we should review this decision. 104696 1 mike 2014-04-29 17:30:58 +0000 This proposal was accepted in principle by the WG on 2014-04-29, subject to approval of detailed wording. The substantive change proposed is to replace the second paragraph of the rules of fn:collation-key() with: The function returns an ·implementation-defined· value with the property that for all pairs of strings ($K1, $K2): (a) collation-key($K1, $C) eq collation-key($K2, $C) if and only if compare($K1, $K2, $C) = 0, and (b) collation-key($K1, $C) lt collation-key($K2, $C) if and only if compare($K1, $K2, $C) < 0 The remaining changes are purely editorial, e.g. changing the summary, deleting the last paragraph of the Notes, adding examples. 104791 2 mike 2014-04-30 08:20:33 +0000 I have anticipated acceptance by updating the master text (the changes are all tagged at="B-bug25446" so they can be rolled back if necessary). 105080 3 josh.spiegel 2014-05-05 21:19:55 +0000 I don't see the change. Did you check in yet? 105092 4 mike 2014-05-05 21:37:57 +0000 I've committed the XML but haven't yet rebuilt the HTML. 105202 5 mike 2014-05-06 16:58:21 +0000 The proposal in comment 1 was accepted, noting also that collation keys should be implementation-dependent rather than implementation-defined.