3664 2006-09-07 07:47:55 +0000 Definition of type promotion complex and inconsistent 2006-10-19 22:51:49 +0000 1 1 1 Unclassified XPath / XQuery / XSLT Formal Semantics 1.0 Candidate Recommendation All All CLOSED FIXED P2 normal --- 1 jens.teubner simeon public-qt-comments oldest_to_newest 11482 0 jens.teubner 2006-09-07 07:47:55 +0000 Section 8.5.1 of the XQuery Formal Semantics contains the judgment prime(t1) can be promoted to prime(t1') prime(t2) can be promoted to prime(t2') --------------------------------------------- (t1 | t2) can be promoted to (t1' | t2') This rule seems unnecessarily complex. The rule t1 can be promoted to t t2 can be promoted to t ----------------------------------------------------- (t1 | t2) can be promoted to t contains the same definition (together with the remaining rules in 8.5.1). Note that in the original inference rule, t1' could be the same type as t2', which is why I replaced it with t in "my" rule. If promotability holds for prime types, it also holds for the original types (hence, my replacement of prime types with their original types). The same (original) definition is also *inconsistent* with a second rule in 8.5.1: prime(t1) can be promoted to prime(t2) quant(t1) <= quant(t2) ----------------------------------------------------------------- t1 can be promoted to t2 Now if I replace t1 with (t1 | t1) and t2 with (t2 | t2) (which are the same types), and use the definition at the beginning of my report, I can "circumvent" the restriction on quantifiers. 12038 1 simeon 2006-09-26 14:14:18 +0000 I've looked at these rules again, and I believe Jens is correct. The last two rules in 8.5.1 are inconsistent and unnecessary complex. The rule that Jens suggest is both simpler and covers the intended semantics. I would recommend to accept the bug and replace the last two inference rules in 8.5.1 by the rule Jens suggests. - Jerome 12052 2 simeon 2006-09-26 15:57:28 +0000 Jens, The XSLT and XML Query Working Groups have looked at your comment and decided to accept your suggested changes. Best regards, - Jerome On behalf of the XSLT and XML Query WGs