Yeah, good invalidation cache (what uP does most everywhere) practice is to remove the cached data on write or delete and cache on read.
On Tue, Mar 4, 2014 at 7:15 AM, Andrew Petro <[email protected]> wrote: > Ah. The layout cache is invalidated at DistributedLayoutManager instance > init, so long as afterPropertiesSet() is called: > > http://goo.gl/2CMW7t > > and afterPropertiesSet() will be called, since it's an InitializingBean > and DLM is being instantiated by Spring: > > http://goo.gl/UPdqTD > > > There goes that hypothesis. Thanks. > > Eric, you still figure saveLayout() should be invalidating the cache, even > though that's not my particular ghost in the machine? > > Andrew > > > > > On 3/4/14, 8:57 AM, Eric Dalquist wrote: > > That should be invalidating the cache but the layout should also be > removed from the cache at both login and logout as well. > > > On Tue, Mar 4, 2014 at 5:17 AM, Andrew Petro <[email protected]> wrote: > >> uPortal developers, >> >> Do you suppose that DLM saveUserLayout() >> >> http://goo.gl/kRxyqz >> >> ought to be invalidating the layout cache so that the saved layout is >> read back from the layout store on next access? >> >> I suspect yes. >> >> I think I'm seeing goofiness where user changes to received DLM fragments >> (as in, deleting them from the personal layout) take immediate effect and >> persist to the database, but *don't* reflect on logout and log-back-in >> without Tomcat restart. I suspect this is because the in-user-session >> cached copy is being properly updated and the changes are being persisted >> to the database, but they're not being pushed into that cache. I hope >> simply invalidating the cache on save is going to be a reasonable solution >> (rather than heroically injecting just the right new data into the cache >> without round-tripping from the layout store). >> >> I hope to have a cogent explanation and a pull request today. But hey, if >> you already know I'm going in the wrong direction on this, do a fellow a >> solid and let me know? :) >> >> Credit due to Tim Levett for the insight that the ghost in the machine >> I'm seeing might be due to this particular cache. >> >> Andrew >> >> >> >> -- >> You are currently subscribed to [email protected] as: >> [email protected] >> To unsubscribe, change settings or access archives, see >> http://www.ja-sig.org/wiki/display/JSG/uportal-dev >> > > -- > > You are currently subscribed to [email protected] as: > [email protected] > > To unsubscribe, change settings or access archives, see > http://www.ja-sig.org/wiki/display/JSG/uportal-dev > > > -- > > You are currently subscribed to [email protected] as: > [email protected] > To unsubscribe, change settings or access archives, see > http://www.ja-sig.org/wiki/display/JSG/uportal-dev > > -- You are currently subscribed to [email protected] as: [email protected] To unsubscribe, change settings or access archives, see http://www.ja-sig.org/wiki/display/JSG/uportal-dev
