>>>> [log in to unmask] 17-01-2005 13:07 >>>
>> Date: Mon, 17 Jan 2005 12:34:07 +0100
>> From: Theo van Veen <[log in to unmask]>
>>
>> I am still in favour of having a default srw indexset containing
all
>> dc, dcterms and bath names. Especially when we are going to avoid
>> things like having gils.title and dc.title and bath.title.
>
>Once more, there is no single "default" CQL context set.  Each
>application chooses its own default.  (And the only exceptional rule
>is that the "cql" set is the default _for relations and modifiers
>only_.)

You're right. I should not have used the word "default".
But having one core  set of commonly used index names from dc,
dctermscore, Bath,  Utility, Cross  Domain, generic, essential etc.
still makes more sense than having to use a  (rather arbitrary looking)
combination from different element sets. I agree that SRW is not
restricted to a certain application area but in the same way as we have
Dublin Core plus application profiles we could have a SRW Core plus
application specific index sets.

The fact that we have this discussion on the list comes from the fact
that there is no other organisation that is currently responsible for
defining the combination of the index sets that we are using. DCMI deals
with element names and not index names. We use the dc index names also
for searching marc fields.
We need to be able to search on encoding schemes (ISBN, DOI, etc) in
QDC records but who will be responsible for those index names?

>
>This is _good_ news for you, Theo.  Although I think your idea is a
>bad one, there is nothing I or anyone else can do to stop you making
>the "theo" set, defined precisely how you wish (e.g. including
>synonyms for all the DC, Bath, etc. indexes) and making that the
>default context set for your application.  Go nuts.
>

The reason for my proposal was that I wanted the opposite.

Theo