Forum OpenACS Development: Re: Clarifying what name means

Collapse
24: Re: Clarifying what name means (response to 23)
Posted by Tom Jackson on 11/05/03 05:58 PM

Can I ask why you are debating what to include in _the_ central table to OpenACS on an IRC channel? Does this have anything to do with an open discussion? No.

Maybe before putting denormalized data into acs_objects, you want to come up with _how_ that information is going to be maintained.

Collapse
25: Re: Clarifying what name means (response to 24)
Posted by Dirk Gomez on 11/05/03 06:29 PM
It's not an IRC channel, it is being discussed on #openacs. You can look up the logs somewhere on this site. I have zero intention on this being a closed or secret discussion.

How is it going to be maintained? Similar to the current mechanisms, you just need to stuff a bit more information into acs_objects. That can be done explicitly by changing the respective -2 or -3 pages or implicitly by adding triggers.

The information to be added is readily available to a package, however currently every package has its own way of storing that data. This is a way of unifying immensely important information - so important that site-wide packages can hardly work without them.

Collapse
26: Re: Clarifying what name means (response to 25)
Posted by Tom Jackson on 11/05/03 07:45 PM
How is it going to be maintained? Similar to the current mechanisms, you just need to stuff a bit more information into acs_objects. That can be done explicitly by changing the respective -2 or -3 pages or implicitly by adding triggers.

So just add two triggers to each object table, or modify every object create/update function with special code.

Collapse
28: Re: Upgrade woes when changing acs_objects (response to 26)
Posted by Dirk Gomez on 11/14/03 10:34 AM
Back in the very beginning of the Sharenet project did we add a table called acs_objects_description that held name, package_id etc.

I don't recall exactly how long it took us to make the various OpenACS packages we were using on the project to play nicely with acs_objects_description, it wasn't much work though. And the effort paid back handsomely because otherwise expensive and/or very complicated queries had become easy to implement and quick to perform.

Timo did a similar change to the AIESEC code. Maybe he can tell us how long it took there - it's not that long ago.

As for the upgrade. Here's an idea Tilmann had during a f2f meeting (f2f narcosis?). Those fields could be added without NOT NULL constraints to allow for a certain upgrade period e. g. a minor release.