[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: while we're restricting things ..

> What is "a number of things"?  The cord message is the only one I
> know about and I already think that should be changed since that is
> _not_ a one message protocol.  For the edit protocol, we've already
> discussed a more symmetric protocol with at least two messages and
> I'd vastly prefer that.  How often does this case come up over the
> "multiple messages in a protocol"?

I think the "one-message protocol" is a distraction.  One could certainly want a message name equivalent to a protocol name because it's, as in the cord example, the "main" message for that protocol.  I think this will be _extremely_ common if the same restriction is placed on cord names as on message names.  You'll have to create a "whiteboard" protocol, and a "whiteboard-whiteboard" cord.

However, I'd still be amenable to such a proposal.  We might want a convention for the single-cord protocol case, though--something like "whiteboard-main" for the cord type, maybe.

On another note, supposing that Ken's proposed restriction is deemed okay, there are a number of related questions:

  Is it legal to have both a "whiteboard-main" message _and_ a
  "whiteboard-main" cord?

  Suppose I have a "whiteboard" protocol and a "whiteboard-sketch" message.
  Is it legal also to have a "whiteboard-sketch" _protocol_?

In both cases, there's no question of ambiguity as far as the computer is concerned.  Whether an identifier names a message, a cord, or a protocol is determined entirely by its context.  However, for the sake of human error and debugging, it might be nice to add a redundant layer of disambiguation.

BTW, where does the #$#mcp message fit in?