[XML-SIG] Re: Generating XML from scratch

[Rich Salz]

> OK, I'll bite.  Which characteristics make DOM un-pythonic?

Quick reply, with some items off the top of my head.

XML says that the order of attributes and namespace nodes doesn't 
matter, just the name and value.  This maps naturally to Python 
dictionary.  On the other hand, the order of an element's children does 
matter.  This maps naturally to a Python list.

Starting from those two basic concepts, think about how simpler many 
things become -- no addBefore, addAfter, etc, just standard Python list 
slices.  Much other stuff can be thrown out.

The element object should have a "resolve_qname" method which takes a 
'foo:bar' qname and returns a (nsuri,localname) tuple.  This removes the 
need for many of the DOM get.../get...NS routines.
    for k,v in curelt.attributes.items():
        (ns,localname) = curelt.qname_resolve(k)
	... now look at all attriubtes, by qname, ns, or localname

And so on.

>>If you are "just" generating XML, then you will probably go faster if 
>>you use things that naturally fit into the python programming idioms.

Don't call complex API's.  Instead set attributes on objects.  That 
seems to be how ElementTree and amara work, for example.  But I think 
that generating XML is not a very hard or interesting problem, and that 
it is very application specific -- i..e, it depends too much on what the 
local object that you are trying to serialize is.  But I'm apparently in 
a real minority here, so don't listen to me.:)

	/r$

-- 
Rich Salz, Chief Security Architect
DataPower Technology                           http://www.datapower.com
XS40 XML Security Gateway   http://www.datapower.com/products/xs40.html
XML Security Overview  http://www.datapower.com/xmldev/xmlsecurity.html