[Asterisk-Users] PBX Console

[Jayson Vantuyl]

On Wed, Apr 23, 2003 at 04:52:56PM -0700, John Todd wrote:
> I'll summarize instead of text-inserting:  I agree that a text based 
> interface for an operator (perhaps with a fancy ANSI or VT100 
> pseudo-graphics look) would be the best solution for some 
> installations.  Keystrokes are much more operator-friendly than a 
> mouseclick.  Access to directory functions are extremely useful, as 
Strictly speaking, text-based interface is not so much a requirement as
text-based input.  A graphical system still potentially has a number of 
advantages.  That said, the overhead may not be worth it where a
pure text-based interface will do.

> would be some ability to rename call groups by the individual user 
> who is accessing the transfer station.  Gastman (or any mouse-driven 
> app, regardless of the underlying OS) is not what I would put in 
> front of an operator that saw a lot of activity.
This is sort of a midrange app.  I can't imagine an operator scaling too
much.  I mean, most companies that are really big end up with a "company
switchboard" to get you to a department, and then a department
"operator" (usually the receptionist) to allow for it to scale, so I
question the amount if traffic we're talking about here.  My question is
simply, where is there a benefit?

With an operator's panel, there is a clear benefit of having a
geographic way to locate the appropriate "button" and fairly dense
labelling of what they do.  In a text-based interface, the labelling
cannot be so dense (you can fit a lot more text on an operators panel
than 80x25) and you can't "reach out and touch" your option.  The
interface, at that point, is relegated to entering numbers, which a good
list will manage quite famously.  The only benefit I see from such a
screen is "call monitoring" (is the line busy, in voicemail, etc), but
there isn't much benefit in the actual routing.

> I'm betting almost all of this can be done the way you want by using 
> the same TCP based interface that gastman uses.  Check it out.  You'd 
> simply be writing your own call manager application on top of the 
> pre-existing API-ish interface, which was (I assume) why that 
> interface was abstracted to a TCP socket with generic data about call 
> states.
After examining it, I can safely say the protocol is the height of
simplicity.  It appears to be plaintext authentication followed by
round-robin interrogation/answer pairs much like POP3 or SMTP.  I highly
suggest an application called tcpflow.  It takes filters like tcpdump
but will either dump the sniffed/decoded stream to files or even to the
console.  As such, you can easily get to work by watching gastman talk
the Asterisk.

I will mention that I'm interested in this because I've already started
one of my own.  I've been using a Python script to mediate between * and
an XMLRPC-enabled application (I'm using XWT but I'm trying to
effectively expose the management API via XMLRPC, not create a new,
proprietary protocol).  When it is done to the point of functioning (and
I can hack out all of the built-in test passwords / build an auth
interface) I'll post something to the list.

Jayson