On Wed, Mar 28, 2007 at 07:33:23AM +0200, Olle E Johansson wrote: > > 28 mar 2007 kl. 01.16 skrev Nicholas Campion: > > > You could create another table in the database that would hold the > > last write times of these tables. Create a separate table with the > > columns "table name" and "mod date", then create or modify the > > wrapper which you use to write information to the RT tables to, > > before they write to the appropriate real time table, update the > > time in the mod access time for that table to the current time. > > Then, when a call comes in, check that table entry and see if it > > needs to be reloaded. > > > > I don't know if that makes this a good idea, but it seems possible. > > > Realtime is not an SQL abstraction layer, it's a storage abstraction > layer. Solving realtime architecture problems > with SQL queries is not a solution. > > Remember that realtime also has drivers for LDAP in the bug tracker > to prove this.
understood. Anyways, in the end there is a caching issue, and one thing that would help here is some means to tell if the cache is up-to-date or not. In terms of abstractions, we could provide a rt_cache_valid() function that is implemented in various ways depending on the backent, and defaulting to "no, sorry" in case there is no supported way to tell for a given backend. cheers luigi _______________________________________________ --Bandwidth and Colocation provided by Easynews.com -- asterisk-dev mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-dev