Addison Phillips wrote:
> 0*1[script] == SHOULD NOT use
> 2*[script] == SHOULD use
That's the KISS variant, therefore I like it. Misha apparently
wanted a more complex solution allowing to mix a default script
with _additional_ expected scripts. I don't see where Misha's
feature is necessary for 3066 compatibility, but it's possible.
> a hypothetical case. Let's say language 'xyz' is listed
> initially with no script information.
That could be a spoken language without script, a case of two
or more undetermined expected scripts, or a very dubious entry.
> A user asks to register 'Cyrl' as the default script,
> produces information in language "xyz-Cyrl" to prove the
> point, and the result is registered.
Better, now it has a script, xyz is processed as xyz-Cyrl.
> Later another user complains that the language is also
> written in 'Latn'.
Bad, the first user and the tag review got it wrong. In the
KISS-variant Latn would be simply added => no more default, all
content tagged as "xyz implicitly xyz-Cyrl" should be updated.
As soon as there's no more default script 3066-compatibility
is impossible. It's one of the two points of RfC 3066bis that
in this case the script should be used / added, a.s.a.p. Here
a bit later than a.s.a.p. because the first user got it wrong,
but the final effect for 3066-compatibility is the same, it's
not possible for this language.
[Misha's variant]
> 1. Does the default script get removed and two expected
scripts created?
One solution, same effect as in the KISS-scenario.
> 2. Does the default script remain and a new expected script
get created?
Possible, Misha's variant allows this. The hard question is
which of the two scripts should be the default. And what's
the purpose of the other script, for info, a hint for taggers,
plausibility tests ?
> 3. Does the default script remain and two expected scripts
get created?
That's not okay, it results in a NOT RECOMMENDED SHOULD. You
could assign the same meaning for this case, but if it's the
same as in (2) you better forbid (3).
> 1. If the primary language has no associated script, you
> SHOULD NOT use a script subtag unless it adds
> distinguishing information for that context.
What's that, a spoken language, pictures of hieroglyphs or
knots, a WAV with Canarian whistles, Zxxx, Zyyy, or Zzzz ?
Your points 2..4 are what I called the KISS-variant.
> 3. If the primary language has two or more associated
> scripts and the content uses one of them, you SHOULD use
> the script subtag (unless it is harmful to do so).
When would it be harmful ? It just doesn't work with RfC 3066
implementations or content. It's more like "unless it's useful
to get any content, especially pre-3066bis content".
> 4. If the primary language has any number of associated
> scripts, but the content uses a different script, you
> SHOULD use the script subtag (unless it is harmful to do
> so).
Same as 3, here it's only something like Brai, not Latn/Cyrl
or Hans/Hant. RfC 3066bis is simply not fully compatible with
RfC 3066, the awful "default script" kludge can only fix this
for your case 2, but not for your cases 3 and 4. Bye, Frank
_______________________________________________
Ltru mailing list
Ltru at lists.ietf.org
https://www1.ietf.org/mailman/listinfo/ltru
Note Well: Messages sent to this mailing list are the opinions of the senders and do not imply endorsement by the IETF.