This is done. I have also changed the 0.10 client package name to be consistent with the one used in common. So, I have moved org.apache.qpid.nclient to org.apache.qpidity. What do you think about this name? Should we keep it or change it? I think that whatever name we go for, we should use it consistently between common and client (and eventually broker).
Arnaud On Mon, 2007-08-06 at 13:41 +0100, Rupert Smith wrote: > I think it would be better to move the selector code to common now. I > never like to see cut and paste code, as there is always the > 'certainty' ;) that the two copies will evolve separately, and be hard > to merge back together again. > > On 06/08/07, Arnaud Simon <[EMAIL PROTECTED]> wrote: > On Mon, 2007-08-06 at 13:19 +0100, Robert Godfrey wrote: > > Hi Arnaud, > > > > In the 0-10 Java Broker (whenever we start on that) I > *definitely* > > expect that we will be having server-side selectors. > > > > We are adding them client side so that we can have JMS > compliance with > > brokers which do not implement server side selectors ( e.g. > our own C > > ++ broker :-) ). > > Broker side message selection is certainly faster than client > side. We > will need to have a negotiation mechanism for finding out if > the broker > provides such a service though. > > > As to XPath and XQuery mesage routing... I think this is a > separate > > issue (and before we say "definitely" I think we should > discuss on the > > list), and is a matter of custom exchanges. > > I used "certainly" with the meaning of "surely = probably" and > not > "definitely" as I would not implement anything without a > prior > discussion. I should have used "probably", sorry for the > confusion. > > >
