& XSLT Philippe Poulard said:
>  Michael Kay wrote:
> >>"The flaw with grammars is that they only allow to constraint
> >>content models in a declarative manner
> >
> >
> > There's nothing wrong with constraints being declarative
> 
>  Except that you will need 3000 tags to express all that people expect
>  and will expect.

Philippe and Michael have different meanings for "declarative".  I think
Philippe means declarative in the sense of the specific semantics of the
operation being labelled. I think Michael means it in the sense of their
being a general label for the whole thing, combined with the properties
being expressed non-procedurally or functionally.

It is the difference between a "declarative" label
  X is an ID attribute scoped to X
and a "declarative" recipe
  X has a unique value for all Xs - not(preceding::*[@X][@X = current()/@X])
Now these may look the same, but they are not in fact the same; because
the advantage of the first is that a semantic label allows re-use: for
example, your software might have a jump table for all IDs to allow faster
link traversal.

So Philipe is saying that there is world of semantics and you cannot make
labels for each of them. Eventually you need to use a recipe. And Michael
is saying you can use a recipe. Not much disagreement actually.

This is where the abstract pattern idea comes in: by providing a simple
parameterization syntax, we can have labels and recipes.

Cheers
Rick



_______________________________________________________________________

XML-DEV is a publicly archived, unmoderated list hosted by OASIS
to support XML implementation and development. To minimize
spam in the archives, you must subscribe before posting.

[Un]Subscribe/change address: http://www.oasis-open.org/mlmanage/
Or unsubscribe: xml-dev-unsubscribe@[...].org
subscribe: xml-dev-subscribe@[...].org
List archive: http://lists.xml.org/archives/xml-dev/
List Guidelines: http://www.oasis-open.org/maillists/guidelines.php