[RC5] Concerning mothballed equipment

Charles P. Wright cpw at unix.asb.com
Tue Jan 20 01:37:30 EST 1998


Martin Brandt wrote:
> 
> >So I thought up a way to get computers like mine up into the
> >distributed.net effort and be quasi-networked. In spite of being a
> >386sx/16, it still has a modem (a 28.8 even, one of the first by PPI).
> >Instead of connecting to an ISP just for some blocks, how about if the
> >client dialed up a central "telephone proxy" and used something like
> >Zmodem or HSLink to send and receive blocks.
> 
> I believe any one of us could do it with a bit of custom software.  There
> are two possible setups for the server side:
> 
> On a machine with a dedicated net connection -> the software on the
> client's
> side would dial in and send it's out-buff file (complete file, even zip'd
> if
> that makes it much smaller), e-mail address, and # of blocks desired.  The
> server side then "edits" an INI file to the e-mail and # of desired blocks
> of the calling system, places the out-buff in the directory, fires up
> rc5des.exe -update, and sends the resulting in-buff back to the
> calling system before hanging up.
> 
> On a dialup net connection machine -> I believe the same kind of thing
> could
> be setup using the regular personal proxy software, but I haven't had any
> reason to use it yet.
> 
> On the client side we run the DOS client (i'd imagine most of these
> machines
> will be setup with DOS) with the -runbuffers option and fire up the custom
> calling software every time it exits.
> 
> So the big question would be, is it worth it?  How many people would use
> such a service, especially if it involved a long distance call every day
> or
> two?  How effective would the addition of these machines be to the effort?
> 
> I could put together DOS versions of the calling software using a FOSSIL
> driver and DSZ or HSLink without much trouble, but I don't have a
> dedicated
> line or internet connection to run it on.  The clients wouldn't need
> anything more than a 2400 baud modem since the buffer files are so
> small...and they could probably even run off a floppy disk with no monitor
> or keyboard.
The calling software really only needs to connect, transmit, and fetch a
pre-made file.  Since these blocks would take so long on these machines
should probably be RC5 only since it takes a while do that contest.  The
buffer could be something like 30 buffers, since this will take about a
week.

The call-up server is a little more difficult, but should be easy to
accomplish w/ Linux.   Have a pproxy on one console, and a dial up
server on the other.  It can have pre-made buffers, and copy the client
program and create a special INI for every time it sends stuff to the
personal proxy.  The personal proxy could be attached to a LAN and send
stuff to a personal proxy on another machine, or have a PPP connection
and send to a main proxy.

The calls shouldn't be more than a few minutes most of which is training
the modems since it should only take 128 seconds (w/o overhead) The
following assumptions were used in this computation:
1 a block takes 256 bits to store in the buff-in.rc5 and buff-out.rc5
There are 30 Blocks in Each
The connection is established at 120 baud (I don't think they ever made
modems slower than this for the PC, the slowest common one was 1200
which would only take 12.8 seconds).

I wouldn't use this because NICs are cheap and it is not too hard to tie
it into my existing LAN with a personal proxy.  However it would be an
interesting endeavor.  

-- 
Charles P. Wright
cpw at unix.asb.com.

--
To unsubcribe, send 'unsubscribe rc5' to majordomo at lists.distributed.net
rc5-digest subscribers replace rc5 with rc5-digest



More information about the rc5 mailing list