Perhaps something more integrated. I.e. an aggregate expression will be part of a query, with an option to return the main object together with an aggregate (more generally - with an option to customize the SELECT clause). E.g. an equivalent of the following EJBQL which already works:
"SELECT a, count(a.paintings) FROM a" Andrus > On Dec 2, 2014, at 7:28 PM, Michael Gentry <[email protected]> wrote: > > Will the AggregateUtils thing I did be helpful? > > > On Tue, Dec 2, 2014 at 6:12 AM, Andrus Adamchik <[email protected]> > wrote: > >> >>> On Dec 2, 2014, at 1:51 PM, Aristedes Maniatis <[email protected]> wrote: >>> >>> On 2/12/2014 7:53pm, Andrus Adamchik wrote: >>>> >>>>> On Dec 1, 2014, at 2:06 PM, Andrus Adamchik <[email protected]> >> wrote: >>>>> >>>>> I think I'll actually create a "stable" branch for 4.0.M2, as new >> changes will keep coming in and we want the release to happen >>>> >>>> Which I just did: >> https://github.com/apache/cayenne/tree/4.0.M2_prerelease >>> >>> Cool. There is a lot of great work in there. >> >> Yep. My app code looks so much cleaner after I migrated it to the new >> APIs. And while still raw, the new cdbimport rules. >> >>> I think our previous naming regime would have this branch called >> STABLE-4.0 if we are ready to now make trunk/head available for new work >> that will not land into 4.0. >> >> I didn't give much thought to the branch naming in this instance, as I am >> planning to merge and delete this branch after the release (leaving the tag >> around of course). So it is a short-lived thing. >> >>> If this really is the end of all new features, then perhaps the next >> release will be 4.0-beta1? >> >> There's more query API work planned. We need to align ProcedureQuery and >> NamedQuery, and create a fluent thing for non-selecting SQLTemplates. >> Though these can be done pretty quickly. >> >> A bigger project is to bring all EJBQLQuery functionality into >> ObjectSelect - multi-result selects, aggregate queries, etc. This is more >> challenging. Whether this stuff should be 4.0 or say 4.1 is debatable. I >> feel there's lots of benefit in a quick completion of 4.0. There are users >> who are only allowed to use "GA" tagged dependencies in their work. And by >> the time we get a "full solution", a few years may pass and it won't be as >> exciting by then :) So I am +1 on the Beta idea. >> >> Andrus
