Re: Specification for XID_Start and XID_Continue

From: Martin v. Löwis (
Date: Wed Aug 15 2007 - 12:40:24 CDT

  • Next message: Petr Tomasek: "Syriac accent marks?"

    > I think it would be useful to add a more detailed description of the
    > derivation; I'll propose that to the editorial committee.

    Looking at the proposed text, it is now much more detailed, giving
    precise category IDs in table 2. I suggest that this is enhanced by
    explaining that the "plus stability extensions" part is formalized
    as "plus stability extensions (Other_ID_Start)", and then the
    set notation becomes [[:L:][:Nl:][:Other_ID_Start:]] - assuming this
    is what is actually meant there.

    As for the reviewer question in 5.1.2 ("Should this change be maded to
    ID_Start and ID_Continue also?"): This isn't really possible, is it?
    a) it would violate identifier stability if the characters were removed
       from ID_Start, and
    b) ID_Start is a computed property, you cannot edit it directly - only
       by changing the underlying subsets. As it is a union, you cannot
       really *remove* from ID_Start/ID_Continue, unless you change the
       categorization of the character to be removed (e.g. to not be a
       letter anymore), or you change the formula by which ID_Start is
       computed (e.g. excluding Lo from ID_Start - which is surely


    This archive was generated by hypermail 2.1.5 : Wed Aug 15 2007 - 12:42:34 CDT