On Thu, 2012-09-06 at 18:53 +0100, Gordon Sim wrote: > On 09/06/2012 06:24 PM, mick wrote: > > It seemed reasonable to me to keep it within the queue code, because > > browse-only-ness is a property of queues, and not of sessions. > > While I agree there is some logic to that, I think the interface for > subscribers to the queue is sufficiently ugly at present that it seems > less neat to pursue that option. At least to me. > > For example having subscribers themselves decide if they acquired the > message then directly calling dequeue() is ugly and necessitates > ugliness if trying to keep everything in the queue. > > > By moving the enforcement of browse-only-ness out of the queue code, it > > seems to me that opens a possibility that some other pathway will > > someday be created -- not through the SessionAdapter code -- that will > > unknowingly violate the browse-only-ness of the Queue. > > Fair point, how about a test on Queue::consume() that throws an error > for any attempt to pass in an acquiring 'consumer'. (The terminology in > this part of the code makes me weep!) > >
Thanks, that makes sense. The terminology made me weep too -- but I increased my meds and now it just doesn't seem to matter. --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
