3260 2006-05-09 11:21:22 +0000 Use of "-" in regular expressions 2008-06-13 19:09:37 +0000 1 1 1 Unclassified XML Schema Datatypes: XSD Part 2 1.1 only PC Windows XP RESOLVED FIXED cluster: regex resolved P1 normal --- 1 mike cmsmcq www-xml-schema-comments oldest_to_newest 9674 0 mike 2006-05-09 11:21:22 +0000 QT approved comment (General comment) in regular expressions defining permitted values for data types, the character "-" is sometimes preceded by "\" and sometimes not. The schema regex syntax, I believe, does not permit the "\". 9746 1 davep 2006-05-11 02:07:16 +0000 (In reply to comment #0) > (General comment) in regular expressions defining permitted values for data > types, the character "-" is sometimes preceded by "\" and sometimes not. The > schema regex syntax, I believe, does not permit the "\". There was a time when there was a question of how to reconcile the use of '-' as a metacharacter in an seRange, and during that time some REs were written with escaped '-' characters. They need to be cleaned up, but they are essentially typos so we did not hold up Last Call to fix them. 20326 2 cmsmcq 2008-05-30 04:05:21 +0000 I believe the premise of the comment is faulty: there is a single-character escape for "-" which can be used whenever "-" itself must be denoted, whether inside a character-class expression or outside it. Consistency and clarity need to be weighed, sometimes against each other. So I believe the correct resolution of this issue is as WORKSFORME. 20463 3 cmsmcq 2008-06-13 19:09:37 +0000 The XML Schema WG discussed this issue at its telcon of 13 June 2008, and agreed that in view of the work we have recently done on regular expressions, which has removed some of the variation in the use of - in the lexical-space regular expressions, this issue should be closed as FIXED. We reminded ourselves that bug 5321 on the EBNF notation used to define the regex language is still open; MK pointed out that some confusion arises from the fact that the grammar notation uses some regex-like constructions, but not always exactly the same regex constructions as the regex language we are defining. The editors expressed a hope that we could reduce the opportunities for confusion by resolving 5321, which is classed editorial. Michael, if you would report this resolution to QT and indicate their acceptance or objection in the usual way, it would be helpful. Thank you.