Enrichments offer a mechanism to asynchronously modify rows. It seems like a natural generalization to offer a similar asynchronous mechanism that can generate or edit rows as well, or generally edit a table.
One application is periodically synchronizing with an external data source, or using a ChatGPT plugin to generate a table from a query.
Another application is to do periodic data validation on a table using multiple rows and reporting back the errors.
The asynchronous jobs, job_ids, and error logging are common to all of these applications, suggesting they could move to datasette proper.
Enrichments offer a mechanism to asynchronously modify rows. It seems like a natural generalization to offer a similar asynchronous mechanism that can generate or edit rows as well, or generally edit a table.
One application is periodically synchronizing with an external data source, or using a ChatGPT plugin to generate a table from a query.
Another application is to do periodic data validation on a table using multiple rows and reporting back the errors.
The asynchronous jobs, job_ids, and error logging are common to all of these applications, suggesting they could move to datasette proper.