Casefolding Sigma (was: Re: IDNAbis PreprocessingDraft)

Michael Everson everson at
Fri Jan 25 14:02:57 CET 2008

At 21:09 +0900 2008-01-25, Martin Duerst wrote:

>  >Canadian Syllabics finals are letters. They are prototypically 
>final consonants and often marks of labialization. One *might* call 
>the labializers "modifiers" but one certainly cannot do so for the 
>finals. Your analysis was based on an opinion of the superficial 
>glyph appearance, not on the actual use of the characters.
>Oh, I understand. For a syllabic script, that makes perfect sense.
>Thanks for helping out.


>These should definitely be allowed in IDNs, they are quite like the 
>Thaana marks,

Um, no. Thaana marks are non-spacing vowel diacritics. (And they are 
obligatory, not optional as in Hebrew and Arabic). The Canadian ones 
are spacing characters.

>except that they (mostly, Sayisi seems to be somewhat of an 
>exception) denote consonants at the end of (otherwise open) 
>syllables rather than vowels after consonants.

Yes, well, that means they are in no way like the Thaana characters.

Get it, Martin? Your analysis was in no way accurate. I have to say 
this harshly. On what did you make this analysis? Casual glyph 
resemblance? That is not good enough.

>Please correct me if I'm wrong. Anyway, I hope you agree with me
>that these need to be available in IDN.

Yes, I do agree.

>U+1424 and U+1426. The names indicate what they might look like,
>but my copy of the book really has nothing there, neither in the
>glyph chart (p. 684) nor in the character list (p. 687), unless
>it's printed in some invisible ink or so.

Oh, my stars. Well, that was certainly missed. Thank you for spotting this.

>Yes. I didn't say that New Tai Lue has a Virama. I just said that they
>are the equivalent of what would be encoded with a consonant and a virama
>if New Tai Lue were encoded using the Indic model, and that following
>the logic of that model, this distinction (virama or not, in a logical
>sense, independent of whether it's encoded explicitly or not), the
>distinction between final and non-final in New Tai Lue should be kept
>in IDNs in the same way as the distinction between consonant and
>consonant+virama is kept in Indic IDNs.
>Again, please correct me if I'm wrong.

You're wrong because you're talking about New Tai Lue as though it 
had a Brahmic encoding. It doesn't. So why bring that into it? You 
don't have to do a glyph analysis of New Tai Lue and compare it to 
Khmer or whatever in order to determine what is in IDN. You need to 
look at New Tai Lue in its own right.

>  >My goodness, this activity can be disheartening.
>For an effort such as IDN, it's very important to have experts
>from all kinds of fields. Your input is therefore very appreciated.
>If everybody on this list would just say things you always had to
>agree with, you wouldn't have to be on this list, or would you?

I hear that the Ethiopic Wordspace is on its way out again. Whose 
bright idea was that, and what was the rationale this time? I think 
forbidding it does nothing but disservice to the user community.
Michael Everson *

More information about the Idna-update mailing list