- From: Rick Jelliffe <ricko@allette.com.au>
- Date: Fri, 20 Jun 1997 15:11:53 +1000
- To: <w3c-sgml-wg@w3.org>, "Tim Bray" <tbray@textuality.com>
I just don't get it. In > http://www.textuality.com/xml/namespace.htm Tim and Andrew say that we need a namespace mechanism because we cannot use URIs. And we cannot use URIs because "URIs are lengthy and require the use of some characters that would be out of place within tags or attributes". But why not? 1) XML is not geared for terseness. Gavin's little compression tests for the shorttag minimisation debate suggest that lengthier names would not increase the effective filesize (compressed) by much. 2) If a URI uses certain characters, surely that is good enough reason to make XML names also allow them. I think you correctly identify the best solution, but then throw it away too soon. I do not think XML should be constrained to have to support arbitrary cut-and-paste between inadequately marked-up documents. An adequately marked-up document, I say, is one that uses unique GIs: URI-like names. If this means that all existing SGML DTDs do not have adequate GIs, then that is not the end of the world. GI uniquifying is trivial and can be done as part of XML normalisation. In OmniMark, for example it only involves altering things like OUTPUT "<%q>%c</%q>" to OUTPUT "<%g(uniq-prefix)%q>%c</%g(uniq-prefix)%q>" Why make an unnessary problem? The problem is poor names, not poor namespaces. Rick Jelliffe
Received on Friday, 20 June 1997 02:01:13 UTC