Good day, all.

I am interested in approaches taken for returning multiple errors from a given request.

I have created a process to generically dump a single SQL statement to multiple files. This process is being used in a data download job that dumps ~20 files per requestor, with up to 50 requestors being handled at the same time.

Needless to say, I can't stop the entire process for a single (recoverable) error on a single requestor. Taking that thought further, if even one requestor is error-free, the process needs to complete. I then need to inform the calling routine of all the errors.

So, I'm thinking of creating some type of generic 'logging' table with a few numeric fields to generically contain some IDs, then maybe an error number field and an error description field and maybe a couple trigger-populated fields (ID, date/time, user...).

I'm thinking of using autonomous transactions (conditionally, with retries for when distributed transactions cause failure)

This is now starting to look/act a lot like my generic debugging tables, so I'm wondering if I'm going down the right path.

Any and all comments/critiques/new ideas/past approaches/links are welcomed!

Thanks,

- Chris