Re: XSLT 1.0 and XSLT 2.0

/ Michael Kay <mike@saxonica.com> was heard to say:
| It seems to me that the same problem is going to affect every task that
| XProc can invoke: XSLT won't be the only thing to exist at more than one
| version. The solution adopted for XSLT, of having two different tasks xslt1
| and xslt2, doesn't really seem the right general solution. Wouldn't it be
| better to have an attribute such as specification-version="1.0|2.0"?

Not always. And, in particular, consider the case of XSLT where the
signatures of the two steps are quite different. We don't have any
sort of provision for that sort of co-constraint.

| It
| might also be appropriate to have attributes that allow the caller to
| indicate which implementations of the specification are preferred and/or
| acceptable as alternatives.

I'm not inclined to try to standardize that at this time. There's
nothing that prevents an implementor from supporting extension
attributes for this purpose. If we see widespread use of this feature
then I think it might make sense to come back and consider
standardizing it.

                                        Be seeing you,
                                          norm

-- 
Norman Walsh <ndw@nwalsh.com> | If you believe the doctors, nothing is
http://nwalsh.com/            | wholesome; if you believe the
                              | theologians, nothing is innocent; if
                              | you believe the military, nothing is
                              | safe.--Lord Salisbury