[DAS] Re: RFC DAS-Headers
Brian Gilman
gilmanb@genome.wi.mit.edu
Mon, 26 Nov 2001 18:33:15 -0500 (EST)
No you get it. I've also been using Axis and really, really like it. I've
just started to put a migration plan into place for this framework. I
stand corrected here and will do the "right" thing and switch parsers
based upon the Qname I'm looking at.
-B
-----------------------
Brian Gilman <gilmanb@genome.wi.mit.edu>
Sr. Software Engineer MIT/Whitehead Inst. Center for Genome Research
One Kendall Square, Bldg. 300 / Cambridge, MA 02139-1561 USA
phone +1 617 252 1069 / fax +1 617 252 1902
On Mon, 26 Nov 2001, Thomas Down wrote:
> On Mon, Nov 26, 2001 at 05:46:15PM -0500, Brian Gilman wrote:
> > Hello,
> >
> > I agree with this. I'm just worried about the poor DAS 1.0
> > implementors who need to understand what kind of XML document they are
> > getting and what doc handler to instantiate. That's all implemtation
> > detail though.
> >
> > If we are dicussing DAS 2.0 we can add the XML schema type,
> > version and name to the SOAP headers. I have to problem with that at all.
>
> I guess that's possible (although not quite the `normal'
> interpretation of SOAP Headers -- they're usually used to
> provide a context for the core message, things like transaction
> IDs). But I still don't really see why it's needed. The schema
> in force (and therefore the appropriate parser to use) should
> be clear from the (qualified) names of the elements you
> recieve.
>
> I can see that, if you're parsing a whole XML document using
> something like SAX, it can be useful to know what's happening
> in advance. But I can't see anyone handling SOAP messages this
> way. The first-generation SOAP toolkits all used DOM to hold
> the complete message before decoding, so it's easy to pass
> sub-trees of the document off to separate parsing code.
> Axis /does/ do event-based parsing of SOAP messages, but it has
> a dedicated framework for delegating the handling of subtrees.
> So again, I don't see any problem with switching parsers based
> on QNames of elements. [Aside: I've been using Axis quite heavily
> for the last few days, and it's pretty nice. The Deserializer
> system is similar to StAX in principle, if not in detail].
>
> I'd strongly suspect that anyone else who does an event-driven
> SOAP toolkit would follow similar patterns.
>
>
> Or am I missing something here?
>
> Thomas.
>