L2/04-222 From: Doug Ewell Date: 2004-06-07 23:35:19 -0700 Subject: Notes on L2/04-205 and L2/04-210 [...] I just discovered these two documents over the weekend and didn't have time to prepare reasonable responses, so I'll just give you the Reader's Digest condensed versions and hope someone can work them into the discussion. If not, I understand and hope I'll have more time next time. L2/04-205 Romanian Proposal for ISO/IEC 6937 Amendment The request for double low-9 quotation mark is reasonable and straightforward. The request for combining comma below needs to be phrased in true ISO 6937 terms. First, ISO 6937 doesn't have combining characters in the Unicode sense; it has virtual non-spacing marks that come before the base character in question. Second, they are only defined for use with certain base characters, which must be specified up front. What ASRO really wants to request is *four* new abstract characters: * capital S with comma below * small s with comma below * capital T with comma below * small t with comma below to be represented with sequences involving this new virtual non-spacing mark. I'm not sure the Romanian keyboard actually is out of compliance with ISO 6937 if it supports characters outside of 6937 -- namely the S and T with comma below -- but that is another matter. L2/04-210 Definition of a Code Position for German Umlauts This proposal to disunify combining umlaut from combining diaeresis should never get off the ground. It would be horribly expensive in terms of both transcoding from legacy charsets and recoding existing Unicode/10646 data. It would destroy normalization stability by not merely changing the decomposition of existing precomposed letters-with-diaeresis, but making their decomposition ambiguous -- does <00E4> (ä) decompose to <0061, 0308> or the new <0061, 0358>? DIN's point about the number of native German speakers (100 million) only highlights the degree of damage that would be done to existing data. Their case for disunifying diaeresis and umlaut, that libraries sort (e.g.) a-umlaut as "ae" and a-diaeresis as "a", seems too weak to justify the cost of this disunification. By introducing an alternate spelling, which is frankly how 99.99 percent of users would see this, this proposal would also add unnecessarily to the "spoofing" problem. If we can't distinguish "Säul" and "Säul" without being told that one has an umlaut and one a diaeresis, how will users know the difference? Thanks, -Doug Ewell Fullerton, California http://users.adelphia.net/~dewell/