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

more notes from richard



>Are cords a package like mcp-negotiate? if yes why isn't it presented as
>2.5.2 the mcp-cord package (and advertised as such in section 1.4) and if
>it's not a package then maybe remove references to packeges in 2.6?

Cords are a package, although not a _required_ package like
mcp-negotiate.  It probably would make more sense for them to be
section 2.5.2, as long as it's made clear that mcp-negotiate is
required and mcp-cord isn't.  Actually, now that I look closely, I
don't understand a couple of things about the current structure of the
document.  Section 2 is called "The MCP Specification"--I thought that
was what the whole document was!  I guess I would make section 2 be
something about "MCP messages", and then create a section 3 "Standard
packages", with subsections for mcp-negotiate and mcp-cord.

Dave, I think you should consider making these changes, but close the
spec if you don't have time to do them.  They'd probably help clarity,
but wouldn't change the meaning of the specification.  (For this
reason, I also don't think you need to wait for everyone's approval if
you make this type of change; in my opinion, it's basically wording
changes.)

>Couldn't a full message name be something else than <package name>-<message
>name>? if the - was replaced by a character not allowed in <ident> and
><package name> this would avoid all ambiguity (and may even simplify the
>parsing.)

I believe we discussed this change before and Jay whined a lot and the
subject got dropped.  I don't like it, but I think we're just going to
have to rely on members of an organization (or a DNS domain, sigh) to
work these conflicts out with each other; perhaps for MCP 3.0 we can
fix this.

--Erik