example of error that validator misses.

I understand that the validator only applies DTD constraints and not semantic
constraints expressed in the prose of the [DTD comments and] specification
prose.

But I thought you still might be interested in an example of a case where
invalid
coding in the page brought me grief and the validator doesn't catch it.

The URI-reference in question is

http://www.xmlconference.org/xmlusa/2002/mondaypm.asp#wai

A bug in the coding of this page is

  <a name="#wai"></a>

Now, there is a semantic error in using a #fragment of #wai and the NAME
'#wai' because the NAME should be 'wai' for the #fragment to be #wai.

Neither Netscape nor Lynx will follow the broken reference, but IE will.

The page is invalid in its own right (without trying to access it by that
URI-reference) because of the requirement that the html:a.name attribute
meet the 'name' syntactic production which requires an alpha character as
the initial character, and hence does not allow a leading hash.  But the
SGML parser doesn't check this, it seems.

Al

Received on Monday, 11 November 2002 13:59:17 UTC