I do NOT support these 2 premises.

In relation with prefixes like "dc" and "bath" Explain can provide information on index sets but it should not be a prerequisite for doing a search. Explain should only explain that for example searching  for "title" in CQL actually means searching for "dc.title" or "batch.title". I would suggest to use Explain to PREVENT the need for prefixes in CQL. In that case one can always do a distributed search for "title" and only those folks, who want to distinct between "dc.title" and "bath.title", ask (before or after) for Explain to see which "kind of title" is being used by a specific server.

Can anyone explain to me the added value of distinction between dc.title and bath.title in CQL. I understand that due to the abstraction in Z39.50 you need to know the attributeset to know what an attribute means. But in CQL you do not need this because of the use of indexes with user-understandable name and so there is also no need to know a prefix. The prefix does not modify the meaning of an index name (I hope) and does not add any information that is meaningfull to the user or the server. The "Punter - SRW-Meister" example of Mike can be applied for prefixes in general. 

The disadvantage of using prefixes in CQL is that it is not possible to do a search in previously unknown databases with usual indexes like title, author and subject. This prevents the concept of "EasyLink" (links, being automatically generated SRU-queries). I can now use a few lines of code to translate OpenURL's to SRU-queries. With previously unknown prefixes this becomes very hard.

In my point of view not supporting Ralph's premises means not supporting prefixes. Or did I misunderstood previous discussions and is everyone already on this track? 

Theo



>>> [log in to unmask] 02-05-02 15:48 >>>
"LeVan,Ralph" wrote:

> No!  The Explain service will say waht prefix is expected!

Ralph, I think that you are holding firm on two premises that I don't think we
yet have consensus on:

1. That Explain will be a prerequisite to doing anything useful in srw.
2. That being able to send the same search (i.e. identical string) to
different systems is not important.

You might be right on both counts, but I think it's time we put these premises
to the test.

I request that others on this list weigh in on these two premises.

--Ray