-
Notifications
You must be signed in to change notification settings - Fork 13
Closed
Description
Currently we cache the list of codegens reported by the backend server in the same tinydb file that we use to cache the transforms. Since this is only updated via user intervention, if a new codegen is added on the server side, it will not be visible by users with "old" cache directories. This seems like a premature optimization (if we retrieve this list once per ServiceX session, it's one simple REST request, where we are going to make many more later) and easily causes user confusion. I would either get rid of the caching entirely, or have a code path where a cache miss for a requested codegen triggers a new lookup with the server for a potentially updated list of codegens before declaring that the requested codegen is not available.
Metadata
Metadata
Assignees
Labels
No labels