[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Comparison of hoffman-idn-reg and jseng-idn-admin

At 07:54 03/04/03 -0800, Paul Hoffman / IMC wrote:

At 6:55 PM +0430 4/3/03, Roozbeh Pournader wrote:


The first two characters are only used in the Arabic languages, and the
third is only used in Persian. The above data is necessary if you want
to distinguish the first two (that are distinguished in Arabic), but avoid
security problems (U+06CC looks *identical* to U+0649 in isolated and
final contextual forms, and *identical* to U+064A in medial and initial

Small aside: Would it be possible to create rules (not necessarily checked by a table approach) that are less restrictive by taking into account the position of the character with respect to other characters? (For non-experts of Arabic: this is not the position in the label, as there are some characters that don't connect to one side, and therefore make e.g. the following character take an initial or isolated form even if it's in the middle or at the end of a word.)I.e. we would have something like:

For final and isolated:


In medial and initial form:


Now, when one registers a label with a U+0649 and someone else goes with a
U+064A, who will own the U+06CC version?

The first person to register either name. I can add (hopefully) "clear language" on this in the next draft.

Okay, this clears things up quite a bit. You definitely have to make that much clearer in your draft. This also makes it understandable that you require all the blockings to be put in a database, because otherwise, it would be difficult to reconstruct them (the exact sequence of registrations would be needed for reconstruction).

But I think such a sequence-dependent structure is difficult
to understand and treat mathematically. That's maybe why
Adam has problems with it. It would be much better if we
had a structure that didn't depend on sequence. Something
along the lines of 'blockings can overlap (and don't belong
to anybody), but mappings cannot overlap'.

Regards, Martin.