ietf-mxcomp
[Top] [All Lists]

Re: suggested new RRtype experiment

2004-05-22 14:49:29


On May 22, 2004, at 3:27 PM, Matt Sergeant wrote:
Most modern XML parsers will croak on this. libxml2 will for example.


Really!? Last year I did an XML parser survey project and all 5 of the parsers I tested had the ability to turn namespace processing off. In fact, I just ran "<foo><a:bar/></foo>" through Xerces with no issue.

Admittedly, libxml2 was not one of them, but I'm having a hard-time seeing this. It claims to support SAX2, and one of the SAX2 features that can be toggled is "http://xml.org/sax/feature/namespace";. And certainly, if libxml2 cannot turn namespace processing off, then I'm sure there are plenty of other parsers that will suffice, such as all the ones written before the Namespaces in XML spec was created.

There's no reason to fear a http URI in the namespace, unless you don't
understand XML, in which case I suggest the griper go learn about XML
instead of griping.

I think you are missing my point (perhaps you are thinking of what Doug is saying). I'm simply looking for ways to keep the XML short. Omitting the namespace declaration may be one of them. A namespace declaration of " xmlns='http://ep.net' is 22 octets, or 4.3% of our precious 512b space. So 2 extensions plus the base could eat up 13% of the payload just in namespace declarations.

Or trust those who do understand XML.

(I still don't think XML is a good idea in DNS TXT records, so this is
just trying to keep the mailing list on a factual level)


I agree: facts are important.

-andy