- From: Martin Duerst <duerst@w3.org>
- Date: Tue, 02 Apr 2002 18:06:07 +0900
- To: www-math@w3.org
- Cc: unicore@unicode.org, ion@ams.org, bnb@ams.org, w3c-i18n-ig@w3.org, asmusf@ix.netcom.com, kenw@sybase.com

Dear MathML WG, This is an errata report from the Unicode consortium's internal technical mailing list. I haven't found these at http://www.w3.org/2001/02/MathML2-errata. I'm acting as the liaison from the W3C I18N WG to the Unicode Consortium, and have copied the I18N IG. For background, the discussion started because there is a note in Unicode 3.2 that variant selectors cannot be used for decomposable characters, to avoid normalization problems, (see http://www.unicode.org/reports/tr28/#13_7_variation_selectors) but there are two cases where this is not respected, which lead to the following proposal: Especially in light of the fact that there is an already-encoded way to express the variation, the UTC should issue a corrigendum removing the variant sequences <2278, FE00> and <2279, FE00> and request that WG2 do the same in a corrigendum to 10646. At 10:42 02/03/29 -0800, Kenneth Whistler wrote: > > See http://www.w3.org/TR/MathML2/variants.html > >Yes it does (have an impact on MathML). >http://www.w3.org/TR/MathML2/cancellations.html > >which *does* deal with the two characters and two variants >in question, and botches those, as well. > >In particular, the slanted cancellation of U+2276 is defined >as <U+2276, U+0338> (correct), but the vertical cancellation >is defined as U+2278 (incorrect). U+2278 is canonically >equivalent to <U+2276, U+0338> -- and what they *should* have >in the table is <U+2276, U+20D2> for the vertical cancellation. > >Same issue for the cancellations of U+2277. >Incidentally, the MathML tables also make the nonsanctioned >extension of all character short identifiers to 5 digits >(defined neither by the Unicode Standard nor Clause 6.5 of >10646-1). I think the standard form is to use either four digits or six digits, is that correct? > > But we should make > > >sure that the deletion of these two variant sequences from our > > >standard would get reflected accurately into MathML, and that the > > >mapping for the not-less-than-or-greater-than-with-vertical-bar, > > >etc. entities get handled correctly. > >I repeat that this needs to be done. > >Any by the way, is *anybody* minding the store over there in >MathML? How do W3C Recommendations get published with blatant >errors in their use of the Unicode Standard and 10646? There is a Math Working Group, and one of its tasks is to take care of errata. And there is the I18N WG/IG, which is doing general review work with respect to Internationalization (which often includes Unicode-related issues), but the mapping details were left to the experts in the Unicode Consortium and the Math Working Group. Regards, Martin.

Received on Tuesday, 2 April 2002 08:10:39 UTC