On Mon, Dec 14, 2009 at 2:33 PM, Andrus Adamchik <[email protected]> wrote: > > On Dec 14, 2009, at 2:17 PM, Michael Gentry wrote: > >> We had a meeting this morning which included a discussion on Hibernate >> vs Cayenne (we = the team that is developing that politically >> sensitive application I mentioned and was trying to do some >> benchmarking). Several of the Hibernate developers, after I showed >> them Cayenne, thought Cayenne looked easier to learn/use (they LOVED >> the idea of the data context), but in the end chose to do the new >> project in Hibernate (the devil they know). A good portion of their >> thought process was that Cayenne 3 isn't officially released yet and >> that Cayenne 2 was too old. (No one uses "beta" code, it was argued.) > > I guess your team should consider that it has a very unique advantage with > Cayenne - a committer who can fix things if they are broken. They won't have > that with Hibernate ;-)
I tried to point that out. :-) I even told them our mailing list was more friendly (at least from what I've heard -- I don't troll the Hibernate lists). It really was a matter of going with what they knew given sensitivity of the application (and the fact that if we botched it, we'd be in The Washington Post). At least we know better than to put a black box over the text in a PDF. :-) > Anyways, I guess we won't be able to fix that DataDomain snapshot cache > contention issue in 3.0, but hopefully we will in 3.1 FWIW, I ran our rather limited benchmark test again earlier and the numbers were MUCH better on Friday. The only thing I can guess is that after we ran the Hibernate test, something happened to slow the DB/network down for an extended period since nothing else changed. Cayenne was still a little slower (but livable) and grew over time (where Hibernate was fairly flat), but it was nothing like on Friday. Also, when I ran Hibernate and Cayenne sequentially instead of in parallel, Cayenne edged it out. >> While I'm on a soapbox ... I think it would be better if 3.1 included >> only a handful of features and could be put out 4-6 months after 3.0. >> Repeat for 3.2 ... That way it doesn't look like it is sitting still >> for to long (perception, I know). > > I am all for it. If 3.1 is just DI/configuration rework (and anything else > squeezed in while this rework is being done), I am personally totally fine > with it. > > Andrus mrg
