Lists Home |
Date Index |
Evan Lenz wrote:
> J. Pietschmann wrote:
>>Nobody in the XSLT
>>community would be hurt if IDs are dropped, many wouldn't
> I would like it if that were true, especially since IDs represent the single
> hole in XSLT 1.0's closure (you can't create IDs in the result tree and thus
> can't do identity transformations). But aren't there whole classes of
> stylesheets that depend on use of the id() function, e.g. Norm Walsh's
> Docbook stylesheets?
Possibly. In the case of DocBook, I believe all ID attributes
have the same name, therefore they could substitute ID/id()
by <xsl:key name="id" match="*" use="id"/> and key('id',$id).
They've faced worse problems in the past. I'm not sure whether
the IDREFS stuff is used, but this can be dealt with too.
For the people bright enough to use a variety of names for
their IDs, they'll face more work, probably defining a key
for each combination and using the union of the key() calls.
Cumbersome if there are a lot of variations, but my personal
view is that they get what they deserve. And in XSLT 2.0, the
complexity can be hidden in a user defined my-id() function,
so apart from the guys writing the adaption layer, nobody's
going to notice much in this case either.