On Wed, 2012-09-26 at 08:08 +0100, Gordon Sim wrote: > On 09/25/2012 10:22 PM, Jakub Scholz wrote: > > Hi Gordon, > > > > The concept in QPID-4339 looks nice. Thanks for your work invested into it. > > > > For our use of Qpid, the priority queues and LVQ are not an issue. You > > do not mention browsing or simultaneous access to the queue by > > multiple receivers, but from the detailed description I assume that > > this will be possible. > > Yes, that should all be possible. > > > The support for messages bigger then one page > > is a bit unfortunate, but maybe for a beginning a configuration of the > > multiplication factor will be sufficient. > > I'll have a think about how spreading a single message across more than > one page might work; or perhaps how extra large 'pages' could be created > when required.
I would guess that keeping a constant page size will make indexing much simpler and faster, so it's probably better to spread big messages across multiple pages. Allowing big pages would penalizing the general case to cater for the special case of oversized messages. Also spreading across multiple pages is a better foundation for eventually handling really huge messages, it gives you a known maximum page size and the ability to page out parts of a really huge message (e.g. everything but the headers) --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
