-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Michael Kay schrieb:
| This actually raises an interesting point that I hadn't thought of before.
| then I think that in Saxon, one index is going to be built for each key
| definition, the indexes will all be searched, and duplicates in the result
| will then be eliminated. So although it does no functional harm to repeat
| the key declaration, there will be a performance cost! This applies
| regardless whether you do xsl:include or xsl:import.

Oh, that's a behaviour I didn't expect. Why don't you look at the key
definition and if there already is a key definition with that name, use
and match attributes, no new index is created? Will you change this in
future releases?

| I haven't actually encountered this problem in practice. I think my advice
| would be to modularize the code so that all access to a particular part of
| the data is channeled through functions and templates in one stylesheet
| module - treat the module as encapsulating that part of the data;
knowledge
| of (and need for) keys on that part of the data will then be confined to
| that module.

wfg,
Jonny.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFHx+REGaEJacf6i/cRAlpCAJ40RhdtDwAkHVdzTPP6kMRBG8xT4QCcDnhS
7CuygnCI2ejFkuLB+U5BT5Q=
=/PIv
-----END PGP SIGNATURE-----