I wrote a bunch of tests for Trond's new basket of quiet commands,
verifying that they all work as expected.

  However, I don't think the semantics are correct.  In particular,
this seems to behave differently from the getq command already
established.

  getq was defined to not be quiet in all cases, but only
uninteresting cases.

  The new commands are defined to never respond under any
circumstance.

  Consider the case of setq:  In almost every case, you can just
assume it makes it, but the error cases can be most interesting.  This
would give us the ability to create a bulk set operation to complement
the bulk get operation.  This should perform quite well, and without
compromising error detection (from CAS or otherwise).

Reply via email to