IMHO, you should not do this at the SQL level, but at the OO level. your "clients" should not change rows, but objects, and the object(s) would then generate sql to update the database, plus any notifications to subscribers.
> -----Original Message----- > From: Rami Saarinen [mailto:[EMAIL PROTECTED] > Sent: Monday, September 06, 2004 12:17 PM > To: [EMAIL PROTECTED] > Subject: [sqlite] How to get the list of rows that were changed during > the last UPDATE SQL-command > > > First of all I have to state that I am not a SQL professional so I may > be asking the silliest thing in the world, but you never learn if you > dont ask. Right? > > To the point: I use the sqlite database in some C++ object > that receives > sql commands from the LAN and executes them. I am trying to create a > kind of poor-man's observer pattern where the SQL object would send > information of changed data to all the subscribers who listen that > particular information (i.e. a single database row). Each row > will have > an unique id. So basically I need to extend the functionality so that > during each UPDATE command I need to get (at least) the > unique id of the > row that was changed. Surely this is common operation for SQL users? > > Should I use triggers? But they seem limited as the only allow the > execution of SQL statements.. so I could have a separate table where > I'd collect the changed information and process the table after the > UPDATE. That sounds kludgey. > > Or I could create a new function for SQL that I could use with the > UPDATE command? Sounds tedious. > > And what about VIEWS? ... Perhaps not what I'd need. > > I'd really appreciate any kind of ideas you might have. > Currently I just > don't know what ( / where) to look for ( / at). > > I am using SQLite 2.8.14 and gcc 3.3.3. > > Thanks! > > -- > Rami Saarinen >
