Hi Rick, Yes, I'm using 2.2 trunk, ~1 week old.
I did try LRS in the past, but as I encountered a bug there I switched away from it. This bug seems to be fixed in trunk, but I could only use LRS for Importing data anyway, as our main DB is MySQL, which lacks proper LRS support with concurrent queries. Thus, I wrote a simple query chunker which loads a chunk of results from the db, provides an iterator, and upon reaching the end automatically loads the next chunk, etc. So, I could try LRS for the purpose of debugging etc, but it should work without them aswell. And Yes! After a quick check, with CopyOnDetach=false, OpenJPA behaves as expected and does not use more than about 30MB of Heap space and most importantly stays in that range. (as long as I detach my entities :) Thanks! I'd still be interested whether it's a bug in the copyOnDetach=true codepath or if it is working as designed that the manager keeps managed copies on detachment? -----Ursprüngliche Nachricht----- Von: Rick Curtis [mailto:curti...@gmail.com] Gesendet: Donnerstag, 11. August 2011 16:11 An: users@openjpa.apache.org Betreff: Re: Possibility to return detached objects from a query? Michael - > While I still wonder about why detaching my entities doesn't reduce my heap memory usage You didn't mention what level of code you are running on, but if it is > 2.0, it might be a bug. Try setting the following property: <property name="openjpa.Compatibility" value="CopyOnDetach=false"/> > also wonder whether there is a possibility for a query to return detached objects in the first place? I'm not aware of one. That doesn't mean the capability doesn't exist, I'm just not aware of it :) > Calling detach() for every object looks to be rather expensive Correct, in some of our performance testing we have found that detach is quite expensive. I put in a property LiteAutoDetach which is in place to quickly detach the entire persistence context, but unfortunately this optimization isn't available for detaching single instances. I know I missed an email you sent to this list a number of days ago where you were asking a very similar question. I don't want to derail this thread, but have you looked at large result sets[1]? The might simplify some of your loading code. [1] http://openjpa.apache.org/builds/latest/docs/manual/manual.html#ref_guide_dbsetup_lrs Thanks, Rick On Thu, Aug 11, 2011 at 2:06 AM, Michael Pflueger <michael.pflue...@sma.de>wrote: > Hi, > > While I still wonder about why detaching my entities doesn't reduce my heap > memory usage, > I also wonder whether there is a possibility for a query to return detached > objects in the first place? > This might be quite efficient for scenarios where you need to read lots of > data but not update it or need any other features of a managed entity. > Calling detach() for every object looks to be rather expensive, and above > method would not only avoid that but could possibly avoid some attachment > overhead aswell. > ___________________________________________________ > > SMA Solar Technology AG > Aufsichtsrat: Guenther Cramer (Vorsitzender) > Vorstand: Juergen Dolle, Roland Grebe, Uwe Hertel, Pierre-Pascal Urbon, > Marko Werner > Handelsregister: Amtsgericht Kassel HRB 3972 > Sitz der Gesellschaft: 34266 Niestetal > USt-ID-Nr. DE 113 08 59 54 > WEEE-Reg.-Nr. DE 95881150 > ___________________________________________________ > > -- *Rick Curtis*
