W3C home > Mailing lists > Public > www-dom-ts@w3.org > October 2001

RE: systemId within DocType Interface

From: Arnold, Curt <Curt.Arnold@hyprotech.com>
Date: Wed, 17 Oct 2001 11:54:37 -0600
Message-ID: <70E215722F6AD511820A000103D141D40AA5E1@thor.aeathtl.com>
To: www-dom-ts@w3.org
If I remember things correctly, I think this would be a good issue for an errata that would clarify what should be returned.  I think all but one parser test (perhaps Crimson) returned the literal
contents of the systemId and the remaining one absolutized the URI.  Definitely the intention of the author is different when they chose to use a relative URI instead of an absolute URI and not
preserving that distinction loses information.  I'll try to confirm the observations on as many parsers as possible.  Probably be good to clarify it in the InfoSet too.

To support either decision, I would prefer to have an explicit <assertURIEquals> element instead of trying to add all the general string processing capabilities to DOM TS ML.  assertURI could look
like:

<assertURIEquals actual="somevar" 
	     scheme="http|file|..." ?
           path="" ?
           host="" ?
           file="" ?
           query=""  ? 
           fragment="" ?
           isAbsolute="true | false" ? >

So for the tests that would accept either absolutized or relative URI's, the test would have been:

<assertURIEquals actual="systemId" file="staff.dtd"/>
Received on Wednesday, 17 October 2001 14:01:46 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 6 April 2009 12:58:45 GMT