[A83] Re: calculator networking - good idea?
[Prev][Next][Index][Thread]
[A83] Re: calculator networking - good idea?
Rather than one calc talking to all the others at one time, sould you not
have a system whereby a calc talks to the hub to get a bried connection
with a particular calc to send a data unit? This allows several transfers
to happen at once, like a telephone switchboard.
At 10:59 PM 1/29/2003 -0600, you wrote:
>Hello all.
>
>I have been toying with an idea for some time now, and I want criticism on
>how practical, plausible, and possible my idea is.
>
>It seems that any calculator communications currently in existence deal with
>an interface between one calculator and one storage device, modem, or
>another calculator. My idea is to create a TI-XX calculator network. Any
>TI calculator or compatible hardware could connect to the network once
>proper software is written. With this network any user on the network could
>transfer variables to and from any other user (with permissions). But,
>multiplayer games and chat applications, etc., could also be developed.
>Devices such as modems, storage devices, and computers with link cables
>could be created or adapted for use on the network, too.
>
>Now, considering that the **very** fastest that the slowest TI-Calcs
>(TI-82,83) can communicate is about 16,000 bps, there will obviously be
>bandwidth constraints. Only one calc would be able to talk at a time, too.
>But I think that the speed could be adequate for most puropses, if we design
>a good packet format and a reliable protocol.
>
>My thoughts for the first, simplest calculator network hardware would be a
>hub with
>four or so 2.5mm jacks. The "red," "white," and "ground" conductors of each
>jack would be respectively connected to the "red," "white," and "ground"
>conductors of the all other jacks. Would one calculator's link port have
>enough output current to power the inputs of three other calculators in such
>an unpowered hub? A powered hub would have simple buffers to assert the
>outputs of each jack.
>
>As for the protocol, each calc on the network could be assigned a unique
>identifier of two bytes, perhaps. Each calc, when it signs on, would ask
>calc $0000 for an identifier. If the operation times out while the calc
>seeks an identifier, it assigns itself the identifier of $0000.
>
>The procedure for sending bytes should be different than the existing
>linkport protocol, which is designed for only two devices. My general idea
>is that there should be a way to detect whether a transfer is already going
>on, and wait until the transfer is done to start the next one. Data would
>be transferred in packets of about 20 bytes or so to prevent a huge transfer
>from hanging up the system.
>
>So, what do you all think? This turned out a little longer than I
>expected...
>
>Jeff
Follow-Ups:
References: