Hey Julian,

I love the system. I have a practical, not technical, question about it. Have you thought about applying this technique to user-triggered agents on databases that might be local replicas, where the server-based log db isn't available? It occured to me that one could tweak the system to save the log docs in that local db, then when they're replicated back up, have some code that moved them into the central repository. But that seemed like it might be too fragile (moving parts and all), not too mention pretty annoying to implement on any db that might become a local replica. I'm wondering if you thought about this while you were designing the system, and what your thoughts were. If you didn't think of it, well, what are your thoughts now? Heh...

Rob -

Well, I guess I figured that any local logging would happen on a local replica of the OpenLog database on the server, so you could just replicate your changes up whenever you wanted. And doing so would write the log entries to the main database, which is the central repository (since the concept is that you would only ever have one OpenLog database on your server, and all or your databases would be logging to it).

You'd probably want to make sure that the replication only sent changes to the server, and didn't pull them down though, because otherwise you could end up pulling a lot of changes down.

- Julian

I hear ya, man. I was just imagining an environment where I didn't have that level of control over WHAT the user replicates. The user may know they want to bring db A with them on the road, so they make a local replica of it. But they won't know to bring db A's log db (or the central log db, in this case).

I know about a hundred ways to try and convince the users to take the log db (sans documents, as you point out), but not many to force them to do so. And one of the most common error triggers I come across in other people's code is that they wrote it without disconnected use in mind (errors out because server or db isn't there). So, in my typically hack-ish mode, I was trying to come up with a way to protect myself from the users (it's my life-long goal to find a way to write code without having to involve users<*smirk*>...).

Rob -

I've been thinking about the problem (and I agree that it's a legitimate one), and I think I have a good option for you. More later...

- Julian

Cool. That's why I asked, 'cause I knew you'd have interesting ideas! Looking forward to seeing what you've come up with...

"(it's my life-long goal to find a way to write code without having to involve users<*smirk*>...)."

Rob, when you figure that out, I will build a shrine for you on my desk and beg to come be your whipping boy because you will rule the world!