"John W Kennedy" wrote: > On Aug 4, 2010, at 8:20 AM, Andreas Stötzner wrote: > > Am 03.08.2010 um 02:47 schrieb David Starner: > >> Fraktur and Antiqua are different writing > >> systems with slightly different orthographies > > > > No. Fraktur and Antiqua are two (of many) different renderings of the Latin > > writing system. > > The two propositions are not mutually exclusive. And it /is/ true that, at > least at some times, Fraktur and Antiqua have had different orthographies.
And it is probably the main reason of the inclusion of "Latf" in ISO 15924, not just because it is a script variant, but really because it defines a distinct orthography, which should be specifiable in BCP 47 language tags. I think you could apply the same rationale on "Hans" and "Hant" as well (not really a different script for the UCS, but distinct orthographies.) Really, "Hans", "Hant", "Latf", "Latg" could have been avoided in ISO 15924, if orthographic variants of the same languages had been encoded in the IANA database for BCP 47, independantly of the effective font style. But for now there's still no formal model for encoding language dialects, so BCP 47 language tags still need to use tags for ISO 3166-1 region codes and for the script variant, when it should just qualify the generic script code (or it could even drop this ISO 15924 code if there was a formal code for the dialect written in a specific orthography: we would also deprecate "Jpan", "Hrkt" in ISO 15924). Orthographic variants would include also: - the various romanization systems (for example Pinyin) and phonetic transcriptions (IPA phonetic, simplified IPA phonology), - the simplified orthographies (e.g. orthographic reforms in French and German), - and some other minor variants (like the vertical presentation for East-Asian scripts, or Boustrophedon presentation for Ancient Greek, if this alters the orientation of characters that had to be encoded differently, and the default mirroring properties are not applicable to the encoded characters in the basic language). For now these dialectal/orthographic variants of written languages can be registered in the IANA database for BCP 47, using codes with at least 5 letters (or with at least 4 letters or digits if there's at least one digit), but ideally the dialectal variant should be encoded as a tag BEFORE the orthographic variant. The font style prefered for each orthographic variant is still left to the rendering system that will apply stylesheets according to the language tag. It should not be invalid to use a fallback style that will ignore the orthographic variants for which there's no font support or no support in the font rendering system or page layout system. Philippe.
