Re: New Canonical Decompositions to Non-Starters

From: Richard Wordingham <richard.wordingham_at_ntlworld.com>
Date: Mon, 18 Feb 2013 20:32:55 +0000

Someone asked off-list if this query were related to a proposal. It is,
to Markus Scherer's L2/12-108. That has been added to the the list of
'stability' guarantees at
http://www.unicode.org/policies/stability_policy.html 'Property
Value Stability', Version 2.1.0+. The redundantly retroactive version
number confused me - I thought it was an earlier stability guarantee.

Unfortunately, the proposed (and implemented) restriction was on the
decomposition mapping property, not on decompositions, so it would be
entirely possible for U+E4567 to have a canonical decomposition to
three non-starters, e.g. <U+0F71, U+0F72, U+E4568>. Such a character
would also foul up the technique of collating FCD strings without
normalisation by extended the set of collating elements, and would be
an even worse nuisance than U+0F73. (That's why I put it in the
loathsome plane!)

Richard.
Received on Mon Feb 18 2013 - 14:34:55 CST

This archive was generated by hypermail 2.2.0 : Mon Feb 18 2013 - 14:34:55 CST