Xinpeng,
On Wed, Jun 5, 2013 at 8:30 PM, Weixinpeng <[email protected]> wrote: > Hi Vince, ,**** > > I am not quite clear about how OUTSIDE_COVERAGE works, if the > database is in charge of returning a list of alternate database with > OUTSIDE_COVERAGE error, does it means the database has to **** > > maintain some information, such as coverage area, of other databases? And > how it works when databases are managed by different companies? > This alternate list is optional, so it's up to the database, business arrangements, and, possibly, regulatory requirements. Because the return value here is very much like what you are proposing, I was exploring whether we should just add a ListDatabase method to the PAWS protocol such that the device just needs to understand one protocol. This would allow: - A database to serve as both listing functionality (if it wants) as well as the spectrum-availability functionality - A server can choose to implement only the ListDatabase functionality In the second case, it would be equivalent to your proposal, just that the protocol messages would be in the same JSONRPC format as the PAWS protocol, rather than the LoST XML format. (If we were to add the method, I think we should remove the listing from OUTSIDE_COVERAGE error). > **** > > If a master moves from USA to UK, can the database that the > master originally connected return available database in UK to master? > That should be allowed. > **** > > ** ** > > Another question is, if a Database Listing server is used, what > interface will be deployed between master and Database Listing server, is > there an existing one, or a new one should be devised? > It should be the same one. This simplifies the device implementation. **** > > **** > > Finally, when HTTP redirect is provided in the protocol, can you > explain it to me when “DbUpdateSpec” be used and when HTTP redirect will be > used, if database wants to provide an alternate datebase? > Great question. There was some thought that HTTP redirect can only return a single URL without additional "attributes", so is less flexible in case we wanted to convey more information about the change. On the other hand, DbUpdateSpec does add a bit more complexity. I think there was also some concern that HTTP redirect is at the outer layer, so does not pass through any hand-shake that might have been required by a WSDB via the INIT_REQ mechanism. So...does the simplicity of HTTP redirect outweigh the benefits? -vince **** > > **** > > Thanks!**** > > -Xinpeng**** > > >
_______________________________________________ paws mailing list [email protected] https://www.ietf.org/mailman/listinfo/paws
