Re: PRI#86 Update

From: Philippe Verdy (
Date: Wed May 10 2006 - 13:35:15 CDT

  • Next message: Asmus Freytag: "Re: some thoughts on encoding Indic fractions etc."

    From: "SADAHIRO Tomoyuki" <>
    > Is a process which adopts this alternative approach to forbid
    > some characters still conformant with UAX#15?
    > I think no; this alternative approach breaks UAX15-C3,
    > as the conformance test includes the corrected mappings.
    > Then, how can the alternative approach, being inconformant,
    > contribute to the stability of Normalized Forms?
    > I assume noway; the forbidding just makes another result
    > that is different from that out of a conformant process.

    Conformance to Unicode requirements depends on its version, regarding the corrigenda. Idon't think that an application that breaks in the cases of characters whose equivalence mappings have been corrected is not conforming, regarding the older version of Unicode on which it was based.

    For this reason, I suggest that "Unicode conforming" should still always include the Unicode number in the conformance labelling, just in case there's a bug corrected later. This way, the conformance can always be checked according to this version (this labelling would be a problem is the corrigenda regarding normative properties did not include a change of at least the minor version number, otherwise applications would need to say "conforming to Unicode 4.0 with corrigenda #xxxx and #xxxx", something very lengthy and really boring comparing to "conforming to Unicode 4.0.1"...

    I would not recommand using the unqualified ambiguous terms "conforming to Unicode" (or the associated logos) in any application, given that conformance to later versions may be compromized if not explicitly tested against the corrections published later (even if this is for obvious errors).

    Regarding the corrigendas published about non-normative properties (such as corrections of the informative representative glyphs), I think that no version switch is necessary (so a font that still uses a glyph looking like the uncorrected representative glyph is still conforming to the same version number, but a font that specifies a later version number should base its glyphs on the corrected represetnative glyph because it has become part of the published standard which integrates now the past corrigendas).

    That's my opinion...


    This archive was generated by hypermail 2.1.5 : Wed May 10 2006 - 13:37:13 CDT