Ross Patterson <A024012@RUTVM1>
Wed, 19 Nov 86 11:24:33 EST
|
>No, this is NOT the wrong list... Imagine a mod to RSCS that allows you
>to specify not one, but MANY destinations (perhaps multiple tags) for a
>spool file. When RSCS stores and forwards this file, it will fork that
>file over the various links it should go, disposing of it when it is
>successful sending it over the last link. Each outgoing copy will only
>including the necessary destination addresses (tags). This feature would
>help the network a lot.
>
>Now if IBM created this feature and supported it, would you say you have
>lost control over your node and its software?
>
> ...
>Does it sound familiar now? Does this mean you have any more or less control?
Phil,
It sure does sound familiar, and no, you wouldn't have lost any
control.
If you changed RSCS to allow multiple destinations for a file,
forking only when neccessary, what you would have is not LISTSERV, but
rather a full implentation of the NJE protocol. It is perfectly valid
for a file to have multiple Network Dataset Headers, each with a
different destination, and in fact such things do happen. JES2 (still
the most complete NJE implementation) will create such animals, and do
just what you ask. RSCSv1 and JES3 immediately split up such a file
when they receive it, rather than pass it along. I'm not sure what
DOS/POWER does (nor do I care ;-) ). The result is that RSCS's
incomplete NJE implmentation kills this goodness. (Maybe the
BITNET/EARN/NETNORTH backbone should be JES2 only ;-) .)
Since RSCSv2 is an NJE-only (no VMB/VMC lines) system, does anyone
know if this has been fixed?
|
|
|