Michael:

> I thought we agreed that WT_NOTFOUND should only be returned by cursor operations – dropping a nonexistent object should probably map the error to ENOENT.

Keith:

> It would be nice if we could somehow map WT_NOTFOUND to an error in all cases outside of the cursor code, I think this will be an easy mistake to make will be hard to find with testing. We could add a macro to non-cursor API-level code, but that seems error-prone as well.

Michael:

> Agree that it would be good to do this consistently. A macro in session and connection methods was all I had come up with.