Hi Ralph,

Can you please describe the problem again?
Did we not decide to remove the concept of database names from SRW (a.k.a.
ZNG) and use different (possibly virtual-) HTTP/SRW Servers to imply the
corresponding databases?

Best regards,
Poul Henrik
mailto:[log in to unmask]

-----Original Message-----
From: LeVan,Ralph [mailto:[log in to unmask]]
Sent: 22. oktober 2001 22:40
To: [log in to unmask]
Subject: Re: Context in RPC calls

As soon as I sent the message, I knew the answer.  That whole context thing
is the antithesis of RPC.  If the data isn't one of the parameters to the
RPC, then it isn't available to the RPC.  There is no such thing as context.

So, does this force us to reconsider whether we need to expose database name
in SRW?  My gut feels is that the answer is "yes", but I don't like it.  I'm
hoping someone out there has a better idea.


>  -----Original Message-----
> From:         LeVan,Ralph
> Sent: Monday, October 22, 2001 4:26 PM
> To:   Z39. 50 Next-Generation Initiative (E-mail)
> Subject:      Context in RPC calls
> So, we're hiding the database name in the URL.  The method
> handling the RPC has no access to URL information.  How is it
> supposed to know what database the search is for?
> Now, I thought I had the answer once.  I was going to define
> a separate servlet for each database, but they would all use
> the same servlet class.  Then, each servlet would get
> initialized with its own parameters and things would be
> great.  But, the tomcat server knows that they are the same
> servlet and only creates one instance of it.  So, I'm back to
> not knowing what database I'm supposed to be searching.
> Now, I can do this myself, if I override the SOAP toolkit's
> doPost method and save the calling URL so that the
> searchRequest() method has access to it, but that seems like a hack.
> Any thoughts on how to handle this?
> Thanks!
> Ralph