MOO-cows Mailing List Archive


Re: Local MCP editing with tf

> >Brandon Gillespie drew these hieroglyphs:
> >> 
> >> This is for those die hards who abhore emacs and still use tf (myself
> >> included).  I dont know if somebody has done this already, so I did it
> >> myself (I know it has been a desired functionality for a while).  The
> >> file: 
> >> 
> >>
> >> 
> >> Will give you MCP editing capability in tinyfugue.  From the header:
> > Now is there a place that actually has an up-to-date spec on MCP?
> > Looking up 'help MCP' on JHM gives me a draft spec that still assumes
> > you're using a hacked server that takes `@@@' as an out-of-band
> > prefix.  This gripe has yet to take root in the do-something-about-it
> > department.
>!mcp is the HTML-ified (and
> commented) version of the spec that everyone agreed on---the original
> is still in!mcp-spec if you
> don't trust my reformat.
> Nobody really remembers why @@@ is in there.  It works well for the
> server->client case, which is good enough to support local editing,
> Connection closed by foreign host.turned in-band.  Works for us, at
> least.
> I'm going to enter a monestary if I don't get my idea of a next draft
> out the door soon.  The line waiting to kill me is getting too long.

I had been collaborating with Brandon to get this code working well
with the 12/95 LambdaCore but looking at this spec it is clear that that core
does not comply with the spec that is available, specifically the value
associated with the name: keyword on mcp type edit is the unquoted
property and whatever characters it may contain including spaces, quotes,
backslashes and whatever else is there. Don't even get me started with what
MOOMail submits for the name: value.

Given this fact I'll continue to bang at this code relying primarily
on the upload: keyword for the data I need to implement intelligent naming of
temporary files (as a hack I instituted a fixed temporary filename in my copy).

Obviously I anxiously await an up to date formal specification for mcp so that
I might bring my core into compliance. So don't add me to the queue of people
in line to kill you Jay, rather include me in the line of people patting you on
the back and offering words of encouragement. :-)

BTW, I would assume that the @@@<data> construct was introduced to prevent the
inclusion of anomalous strings in the lines of data (pages, events, messages,
etc.). This convention and the associated * (to indicate multiple lines) are
not adhered to by the aforementioned LamdaCore either but it appears that this
may be an area of volatility in the spec and another reason I'm waiting for an
updated mcp specification before working on the in DB code.



-- | |
 HijinxMOO (interactive) ==> telnet://
 HijinxMOO (www service) ==>


Home | Subject Index | Thread Index