Hi all,

>From my perspective, adding a NamespaceContext parameter to the
startDocument call seems to make sense.  If it weren't for the fact that
we've already resolved to make one significant change to XNI (adding
methods to the XMLComponent interface), it might not be justifiable.  But,
in the circumstances, I'd also like to see this make its way into the
forthcoming Xerces 2.2.0.

And it's not entirely surprising that we're still finding places where XNI
can be refined; after all, the work that folks like Elena and Sandy are
doing marks one of the first times anyone's really tried to shake Xerces
out for performance.  And Xerces's performance will certainly be affected
by XNI.

But Alex's concern about this being a completely backwards-incompatible
change is absolutely well-founded.  As a community, it behooves us to keep
XNI as constant as possible so that people will have confidence
implementing it; an API is no good if it never becomes finalized.

Since there aren't all that many nooks and crannies in Xerces left to
explore, and since there certainly exist other important applications like
XPP and Neko that are written to--and thus help to validate--XNI, it seems
to me imperative that we set some date soon by which we'll declare XNI 1.0
golden.  We may still want to version it, of course, but then we'll be in
the xsame boat as every other proprietor of a stable API and we'll have to
pick from one of the well-worn solutions.

So could I propose that we set end-of-year as the absolute cut-off for
further XNI (1.0) changes?  If we agree earlier that there are no more
changes which will need to be made, then so much the better.

Does this seem flexiblle enough to other people in the community?  Alex,
does this help give you confidence that soon there'll be a stable target to
program against?

Cheers,
Neil
Neil Graham
XML Parser Development
IBM Toronto Lab
Phone:  905-413-3519, T/L 969-3519
E-mail:  [EMAIL PROTECTED]



---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to