Terrell Russell

2 seems good and correct for backwards compatibility. and 1 seems correct moving forward.

need more tests for the 'bad' cases

I see logic for >4.3.2 for user, but not for group - don't we need to return the old `CAT_INVALID_USER` error to a pre432 or undeclaredversion client?

If merged changes are already in and have caused issues, outside of this PR... then a new PR should be created to clean up any existing problems. If those merged...

I’d say required… except that we’ve already shipped this for a while?

This all seems correct to me. `-S`, and we deterministically work with either the full hierarchy or the leaf, or the (`-n`) replica number. More than one being defined should...

Okay - yes - we should not be depending on this function sequence error as a 'valid' return.

I believe with 4.2.9+, the second and third concurrent `irepl` call in this scenario will be blocked and return immediately since that data object is logically locked. So the result...

I'll mark as duplicate, but leave open for now so we can close them at the same time (in the same milestone).

`iput -f` doesn't currently honor the default resource as expected/defined in the client and server configurations?