On Jun 1, 2010, at 5:00 PM, Kevan Miller wrote: > > On Jun 1, 2010, at 7:24 PM, David Jencks wrote: > >>> Side note, IMO, not supporting it is a failure. >> >> I agree, we would be working around a long-standing bug in geronimo 2.2. I >> was really glad to fix it in trunk and then fix openejb. >> >> To fix it in geronimo, we need an InitialContextFactoryBuilder modeled on >> the aries one, but using gbeans instead of osgi services. This would only >> take an hour or two to write. However it may cause a fair amount of >> upheaval to get it properly installed and configured. I don't have a lot of >> time to spend on this right now, maybe I should try writing the ICFB and we >> can see how hard it is to get it to work. > > Ya. I think we all agree in principal. It's the "fair amount of upheaval" > that I'm worried about... I haven't seen this causing issues for users. So, > haven't been particularly motivated to fix. Esp since it's fixed in G > 3.0/OEJB 3.2 and likely to cause issues on the G 2.2.1 codebase... Seems like > we may be as likely to create new problems as fix old ones...
That's what I thought at first, but we've already experienced significant problems from not fixing geronimo 2.2.... dblevins very sensibly ported the openejb fix to the openejb branch and broke g, and you spent a bunch of time before hunting me down..... I'm not sure we can reliably prevent fixing this bug in openejb branch forever :-) Maybe we should try fixing g and see how hard it is. david jencks > > --kevan
