More code dealing with instance and global parameter semantics. I finallyhave a reasonable model of inheritance and semantics implemented, notperfect from a language-theoretic point of view, but works well withouthaving to rewrite every bit of parameter-handling code in acs-core.
added security check to notifications sweep query - without this check and with no mechanism to remove requests for users who are no longer authorized, notifications still go out. not good.
Fixed up the apm parameter stuff to the point where I can at leastregister and unregister, and cache values of, global parameters. There'ssome really funky code here, I also took the time to have the unregistercode use the unregister API in PL/[pg]SQL rather than roll its own.
Victor had made an upgrade script for version d4, and committed it beforeI got around to adding and committing mine. So now my tcl upgrade commitsand my datamodel ones aren't synchronized :( But they're dev versions sono harm should result.
If I'd known global parameters would be such a PITA, would I have TIP'd todo them?Turns out we must make a parameter value when a parameter is registered, elseall sorts of stuff dies. This is for PG. Oracle will come soon.
Changed primary key constraints to meet our naming conventions, becausethe slightly non-standard name caused my pretty "no errors" page for thispackage to turn red :(.