On Tuesday, 03/26/2002 at 04:57 PST, Shane Curcuru
<[EMAIL PROTECTED]> wrote:
> I presume you're only talking about the custom incremental support that
> we enable in specific situations when we detect we're using Xerces (1.x
> or 2.x), correct?

Yes... although as we move toward using schema-derived type information,
we're going to become more dependent upon APIs that Xerces1 doesn't
support.

> Immaterial of this decision, Xalan will still work by default (i.e.
> both compile and run) with any JAXP-compliant parser, either Xerces
> 1.x, 2.x, Crimson, or others, correct?  The only thing we would lose
> would be the special incremental mode we allow when using Xerces
> directly.

Requesting incremental construction from Xerces1 would fall back on
IncrementalSAXSource_Filter, as we do for incremental handling of other SAX
streams. The main effect on users would be that we would now require
Xerces2 at compile time.

(Which may or may not be a reasonable assumption to make, so there may be a
long-term argument for using reflection here. But I really hope not...)

> Also: is this just a maintenance divot you'd like to simplify

Mostly. I'm mucking about with that code right now in the process of
experimenting with XNI support, and the reflection stuff does impact
readability.

> I'd vote to keep Xerces 1 support for another release or two

I don't feel very strongly on this, so I don't object; just wanted to get a
sense of whether there were likely to be anguished screams when we do
eventually pull the Xerces1 special-case.


Reply via email to