Article:		XML Data Bindings in Python, Part 2
Subject:		Retaining sequence information in gnosis.xml.objectify
Date:		2003-07-09 00:18:44
From:		David Mertz
I thank Uche for doing an excellent job covering my library--heck, probably better than I've done writing about it myself. I want to let interested readers know that I am experimenting with retaining complete sequence information in gnosis.xml.objectify (Uche's spam/eggs example shows where I lose structural information). The motivation is simply that I am working on a little project where I need it myself. So look forward to that, in what will probably be called Gnosis Utilities 1.1. But I haven't yet finalized the interface for this extra information (probably it will be accessed wholly with utility functions, since I do not want to pollute the namespace of a "py_obj" with anything that could conflict with a child tagname in the XML). One reason I have held off on doing this is that I still don't feel comfortable with automating re-serialization. There's a slippery slope between the one thing and the other. Since my objects are so "close-to-the-metal" Pythonic constructs, you can do a lot of flexible things with them. More so than those in ElementTree, PyRXP, or generateDS. The drawback is that many of those flexible things produce objects that have no obvious or straightforward serialization to XML. I've resisted the temptation to guess too much on what users want for a serialization in fuzzy cases. Of course, another issue is that I want to keep the EXPAT version of the parser very memory-friendly. Adding too much metadata about what objects are doing weakens that goal. DOM is obese from the start, so I don't care about it for users of the (default) DOM parser in gnosis.xml.objectify. David Mertz

Previous Message

Next Message