That's a nice addition. We get the benefits of access to session
commands without a bunch of new quad-foos.
I don't really care one way or the other about providing access to
usercmds. One the one hand, it seems self-consistent to allow this
case. On the other hand, it's another extension to IBM/ISO behavior.
I personally don't have an application for `⍎ ']usercmd'`.
That said, I think it'd be nice if errors propagated through the
execute. For example,
⍎')foo'
doesn't complain about a BAD COMMAND, nor is ⎕EM set.