- From: Frank Ellermann <Frank.Ellermann@t-online.de>
- Date: Thu, 21 Aug 2003 11:24:14 +0200
- To: Terje Bless <link@pobox.com>
- Cc: www-validator@w3.org
Terje Bless wrote: > the current beta version will catch this problem Good, somehow the old behaviour went against my intuition, ;-) BTW, the test suite /dev/tests/ is fascinating. Maybe mention it on a more prominent page like /docs/ or /docs/users.html (?) I've found an interesting difference between the W3C and the WDG validators: In one of my documents I've "included" some symbolic MathML character entity references. This worked with both validators. I've then changed the URL of the entity sets to the corresponding validator.w3.org/sgml-lib files, because I use this directory in all DOCTYPE elements. I've no idea why some entity sets exist three times there, e.g. /sgml-lib/UPD-MathML2-20021015/isotech.ent /sgml-lib/UPD-MathML2-20021015/iso8879/isotech.ent /sgml-lib/UPD-MathML2-20021015/iso9573-13/isotech.ent The latter is apparently the file used in the MathML DTD. But because I only wanted a quick hack defining some entitities in an XHTML transitional document, I first tested the version in /sgml-lib/UPD-MathML2-20021015/isotech.ent The W3C validator accepted this. But the WDG validator didn't like the <!ENTITY Ƶ "𝕃"> in this file. I don't need Ƶ in my document and use again the "better" version /sgml-lib/UPD-MathML2-20021015/iso9573-13/isotech.ent For some time the W3C validator did not more accept character references like € - now this works again (resulting in a warning instead of an error). Too late, I finally understood the difference between € and byte 128 in a windows-1252 document (after two years... ouch ;-) But if you want a really weird test case with windows-1252 characters and MathML 2.0 entitities in a "compatible" XHTML document rip <URL:http://purl.net/xyzzy/ibm850.htm> Bye, Frank P.S.: "compatible" ~ 1.1 + attributes align and name... <gd&r>
Received on Friday, 22 August 2003 19:17:01 UTC