Re: multiple trigger programs for many data files - is there a better methodology?

Have you written a trigger program, or read the documentation on trigger
programs?

Think about this. You say perhaps you just want file name and keys. When
you declare a program as a trigger program it automatically receives
certain parameters when the trigger is activated. Among them are the
table name, a before image and an after image. An externally defined data
structure does a fine job of mapping the before image and the after image.
The trick is using one program to figure out the keys for 20 different
files. Do they have the same layout? I don't use a mediator. I do not
believe a mediator would simplify this. I do think that if you are early
on in the process it may help you implement new versions of the trigger
program until you iron out the bugs. A mediator basically would read the
table and say if I am monitoring this table then use this program to
process it and pass all the parameters again. You still have to determine
what columns are the keys for that table and where to record that.

That is if you want to process the data later. After all, you already
have the data that was changed. You could use the trigger program to just
output the 'after' image to a table. Imagine this layout:
- Table
- Timestamp
- Add/Change/Delete
- 'After' image

The problem with that is how do you sort that by Table, 'keys' and then
timestamp descending (to find the last update to a particular key).



Rob Berendt