Hi, On 9 January 2014 05:57, Felix Meschberger <[email protected]> wrote: > Hi > > Am 08.01.2014 um 13:12 schrieb Ian Boston <[email protected]>: > >> Hi, >> I am seeing a thread leak in Sling Event 3.2.0 which is fixed in the >> TimedEventSender in 3.3.0, however Sling Event 3.2.0 depends on later >> versions of Commons Scheduler which has caused some issues upgrading. >> Class cast exceptions and some strange behaviour elsewhere. I suspect >> it can be fixed by updating enough bundles, but that is something I >> would like to avoid. > > I agree with avoiding. But I fear this goes deeper :-( > > Can you elaborate on the Class Cast Exceptions and strange behaviours ? > Ultimately, I think we should fix these !
The problems encountered are not in the Sling code base, and are fixed in later versions of the code that I am working with which uses Event 3.3.0. I can't upgrade due to deadlines. > > Event 3.3.0 seems to depend on Scheduler 2.4.0 whereas Event 3.2.0 dependet > on Scheduler 2.1.0. > >> >> Would anyone have an objection if I backported the TimedEventSender >> from the tip of Trunk and did a maintenance release of Sling Event >> 3.2.0 (ie 3.2.1) with the same package dependency footprint as 3.2.0 ? > > You mean a 3.2.2 branch off the 3.2.0 tag ? yes > > That would be a precedent which I am not comfortable with. me neither, we don't want to have branches off every tag to maintain. (I prefer finding and fixing the root cause for your problem). But if it fixes your problem faster than finding the root cause (yet, we must not forget finding the root cause). > >> >> If thats not ok, I can patch the bundle locally. >> >> Best Regards >> Ian >> >> BTW, I haven't checked if its possible to keep the package >> dependencies static, but I think it will be. >> >> BTW2, I think TimedEventSender is the source of the thread leak.... >> but I cant be certain until I patch and deploy the code. > > You might want to do that on a private copy/checkout/patch before going the > Sling branch/release route, anyway ? I think this makes sense. Seeing as this is a special 1 off case, I'll do that. Thanks Ian > > Regards > Felix >
