Re: xsl-for-each

[Joseph Kesselman]

> > XSLT 2.0 allows Temporary Trees (new name for RTFs) to be treated directly
> > as nodesets, but we can't do that in a fully compliant 1.0 implementation.
>
> Would it be possible to enable by some configuration option a "quirks mode"
> where it is possible to use an RTF as a node-set?

Speaking only for myself: I'd really rather NOT see a quirks mode in Xalan. That's precisely what the extension mechanism was created to address, and it should be allowed to do so. The point of standards is that they _ARE_ standards, and you should be very explicit about when you're stepping outside their scope.

On the other hand.... at some point we'll want an XSLT 2.0 prototype mode (known nonstandard, known nonportable, known higher risk both because new code and because the spec is still evolving), and making the temporary-tree-as-nodeset behavior work there would be entirely reasonable.

______________________________________
Joe Kesselman / IBM Research