- From: Tim Bray <tbray@textuality.com>
- Date: Tue, 27 May 1997 08:19:13 -0700
- To: w3c-sgml-wg@w3.org
I am *not* arguing the general position that we should not improve the spec's handling of conformance and data transmission. Having said that, at least two of Henry's questions are easy. At 03:58 PM 5/27/97 BST, Henry S. Thompson wrote: >1) If a dtd declares an attlist in required markup, and an element >appears in the instance with an additional attribute, is the doc't well- >formed? Trivially yes. The definition of well-formedness does not depend in any way on the presence, absence, or content of attlist declarations. It does require that if they are present they be syntactically correct. >2) If an start tag appears for both an empty and a non-empty element, is >the doc't well-formed? Trivially yes, as long as the empty element also has an end tag. Check the grammar. XML *validity* requires that elements declared empty use the sole-tag syntax (why?). Uh, are you really asking whether you can have both a start tag and empty tag for the same element type? Still yes. >3) Are Names which are 'matched' folded up before they are delivered to >an application, e.g. gis, attr names, and in particular, PI names? It is explicit that case is not significant in [please let's use the right terminology] element types, attribute names, and PI targets. There is currently an acknowledged hole in the spec in that it doesn't say enough about what gets passed to the app. However, given the current language about case-insensitivity, a processor that didn't fold types, attr names, and pi targets would clearly be in violation of the spirit of the spec. - Tim
Received on Tuesday, 27 May 1997 11:20:56 UTC