"LeVan,Ralph" wrote:

> But if I define index "pooh" as being use=1016, structure=6, ... is there
> any problem?

I'm unclear on the current thinking of how explicitly index are defined in terms
of Z39.50 attributes, because we did some serious decoupling at the July
meeting,. For example, the bath indexes aren't tied to explicit attributes
anymore, they just point to the profile.

And take dc.titleWord, for example.   Certainly there is no prescribed use
attribute (since there never has been agreement on a Z39.50 use attribute for dc
elements). So this means that one server supporting dc.titleWord could say that
it corresponds to bib-1 type= use, value 4 another to bib-1 type=use, value
1097, and another XD type=accessPoint value =1.  Right?

And the structure? bib-1 type="structure" value =1, or utility
type=format/structure value = (all these words) (any of these words) ??

And if we say that  indexes are defined in terms of attributes, then which
attributes? Use and structure? (We've decided that for example truncation would
be represented within the cql syntax) and if those two, does that lock us into

I know it's a bit late for these questions; they've all been asked before but
never answered.

 I think before we go further (or at least very soon, and before version 1.0 is
set) we need a paper added as a normative spec that say what an index is.