On Mon, Dec 16, 2002 at 09:40:59AM -0800, Ovidiu Predescu wrote: > Great job, Michael! I'll take a look at it and check it in > appropriately.
No Problem :) > > One thing I think we can change to make things simpler is to disallow > customizing the time-to-live on a continuation instance basis. This > would force continuations to be expired starting from the top of the > tree, making the expiring algorithm a lot simpler. Changing TTL on a > continuation basis does not seem to give us a big benefit compared to > the (computational) complexity of the algorithm. It should be easy to modify expire strategy as required. > > I'll check out your implementation in more detail and see what would be > the implications of the above. Ok, thanks for taking the time. Best Regards, Michael Melhem > > Best regards, > Ovidiu > > On Monday, Dec 16, 2002, at 04:46 US/Pacific, Michael Melhem wrote: > > >Hi all, > > > >FYI, bugzilla Patch [Bug 1538] implements "Expire Continuations" using > >the cornerstone scheduler to periodically (as defined in the xconf) > >clean > >up expired flow continuations. > > > >I decided to use the cornerstone-scheduler because the avalon-monitor > >was > >not appropriate, and I wanted to use a standard avalon component (and > >not have to write my own scheduler;). > > > >The continuations "short-hand" in the roles file has been renamed > >continuations-manager because it is infact ContiustionsManagerImp that > >is being defined here. So xconf becomes: > > > ><continuations-manager logger="flow" time-to-live="3600000"> > > <expirations-check type="periodic"> > > <offset>180000</offset> > > <period>180000</period> > > </expirations-check> > ></continuations-manager> > > > >The above configures the continuations-manager, setting "1 hour" as the > >default time-to-live for contunuations, and setting 30 mins as the > >interval > >between expire checks. Note: all durations are specified in > >milliseconds. > > > >So, If someone can review this patch (and perhaps apply it) then that > >would be great!! ;) > > > > > >Points Of Interest:- > > > >1. Im using what is called atomic-subtree algorithm to expire the > >continuations. (See > >http://marc.theaimsgroup.com/?l=xml-cocoon-dev&m=103893658515510&w=2) > >for explanation of this). It wouldnt be difficult to configure the > >manager to use a different expire algorithm (if that was deemed > >necessary). > > > >2. Currently The latest cut of cornerstone does not appear to work > >with the > >lastest excalibur-thread*.jars. Im chasing this up to get this fixed. > > > >3. As far I can tell the current ECM in Cocoon does not handle > >non-component > >classes (such as the latest cornerstone-scheduler). I have worked > >around this, > >but this issue needs to be addressed. Perhaps we need to update the ECM > >version in Cocoon? > > > >Best Regards, > >Michael Melhem > > > > > >--------------------------------------------------------------------- > >To unsubscribe, e-mail: [EMAIL PROTECTED] > >For additional commands, email: [EMAIL PROTECTED] > > > --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, email: [EMAIL PROTECTED]
