I thought I would pipe-in on this thread and inquire about these changes so that I can consider them for the .NET implementation that I'm trying to kick-off...
From the .NET perspective it looks like it may well be possible to
have an incremental parsing model similar to Axiom. While I do not know of many alternatives to the core XML library provided by the framework, removing the XML infoset data from the model will make other functionality - basically, anything that you want to do with the raw XML (XPath, etc.) - much more tedious than necessary. I can try to explain more, if any one would like more details. Again, this is based on the core XML library; I haven't researched to see if there are other, open source libraries available. Utilizing the core library would allow for both a lightweight, stream-based API (should the Axiom/OM-based model be abandoned) as well as an Axiom-like implementation. My only concern with the Axiom-like implementation is that it may not be all that lightweight once the entire document has been parsed (but, wouldn't that be the case with the Java implementation as well?) since, as I understand things now, the resulting model will extend the core DOM implementation..but I may be worried over nothing... Since I have yet to write any code (much of my work thus far has been running tests, stepping through code, researching libraries, etc.), is there a consensus as to the direction of the .NET implementation? I realize some of this post may be vague, at best, for those who do not have a background or experience in the .NET world. Please feel free to ask questions as I am more than willing to have a dialog on anything and everything. Thanks, Harris On 10/17/06, James M Snell <[EMAIL PROTECTED]> wrote:
So we've been using Axiom for a while now and I think we all agree that there are definite pros and cons. What I'm wondering at this point is whether or not we commit to using it long term. I've been going through the code a bit this week trying to get an idea of what it would take to build an impl of the core API directly on top of StAX while preserving the incremental parsing model and xml infoset. It should only take a couple of weeks to do and would likely duplicate much of the function currently in the Axiom linked-list (llom) impl but it would also give us some freedoms to do more interesting things with extensions, a streaming Atom parser and writer API and a jsp taglib I've been thinking about for building out Atom documents using JSP. At this point, so long as we maintain the incremental parse model, preserve the lightweight memory footprint, and can still support xml sigs and encryption, then I'm fine with moving away from Axiom. Like I said, I could likely have something implemented in about two weeks. Thoughts? - James
