L2/99-339
October 22, 1999

Subject:  ISO 8859-7


I sounds like we should discuss this at the meeting, perhaps in the L2
section.

"Mark E. Davis" wrote:

Coding a new character that looks the same as a another punctuation mark
is questionable. Except for a few cases that slipped through, Unicode
does not distinguish between punctuation or accents that have the same
appearance but different usages in different environments or languages.
We don't have a French dieresis and a German umlaut, an English decimal
point and abbreviation point and period, etc. That is the reason for
making them canonically equivalent when they have slipped through.

On canonical equivalence of these characters, I am afraid the horses
have long left the barn door; even if we wanted to, Unicode 3.0 is out,
and we have guaranteed to everyone that the normalization will not
change. That guarantee is more important that round-tripping any
particular characters, or even fixing any apparent mistakes. (Moreover,
we don't guarantee round-tripping to every new or modified standard; nor
do we guarantee round-tripping of normalized data.)

If we really wanted to have normalized data round trip to the proposed
8859-7, we would have to introduce a separate, new Greek semicolon.

We should all vote no on the new 8859-7.

Mark

Karlsson Kent - keka wrote:


Hi!

   ISO 8859-7 is now being balloted.  In the revision there is
both SEMICOLON (of course), but now also GREEK QUESTION MARK.
These are currently canonically equivalent in Unicode 3.0.
Should these still be canonically equivalent, or just compatibility
equivalent?  This may be something you need to discuss during
the UTC/L2 meeting.  Has there been any Greek opinion on this
matter before?

   Kind regards
   /kent k