Text 8131, 66 rader
Skriven 2016-03-09 00:49:11 av Stephen Hurd (110.fido-ftscpubl)
   Kommentar till text 8130 av mark lewis (1:3634/12.0)
Ärende: FSP-1040.001 draft 1
============================
  Re: FSP-1040.001 draft 1
  By: mark lewis to Stephen Hurd on Tue Mar 08 2016 02:22 pm

 >  SH> There is no documentation I could find on how to set the cost
 >  SH> value of a packet...
 >
 > i don't see anything about a cost field in PKT headers so i'm unsure what
 > you  mean byt the "cost value of a packet"...

Whoops, not sure what I was thinking with this reply.  I'm off my rocker
apparently.

 >  SH> There's not even any documentation of what the cost represents.
 >
 > all i see is "in lowest unit of originator's currency" and the only thing i
 > can fathom that would be related to that is the cost of transporting that
 > single message from the current system to the next system... especially in
 > light of this all having been developed using POTS connections which were
 > not cheap...  it was a feature whereby the operator could specify in their
 > mail events that  messages with a cost <= X could be transmitted during
 > eventA and messages with  a cost value greater than X would be held for
 > another event which would qualify them for transmission...

A reasonable assumption, but the FTSC is supposed to document common practice.
The common practice here is to set the field to zero.  Documenting what it's
not used for doesn't seem in keeping with the mandate of an FTS.

 >  SH> I could add that it's the expected cost the sender bears for
 >  SH> delivery of the packet,
 >
 > yes...
 >
 >  SH> but it's not clear if it's sender costs, receiver costs,
 >
 > the sender has no clue what the receiver's cost is...

That's not universally true.  Back In The Day, it wasn't unusual for sysops to
negotiate extensively and cost share.  The cost field would be set to the total
cost.

Also, the "cost" was often variable depending on how/when it was incurred.
Even the scenario you put forward where certain times would allow different
costs was because the cost was lower at certain time.  The cost was therefore
*not* the cost to deliver the message, but the cost to deliver the message
during the high-rate period.

If I pack a message a 2:00am, the cost field is unlikely to be different than
if I pack it at noon.

 > it always seemed obvious to me because of operator set values in mailers and
 > bbs packages and those being specifically referenced to be the cost of the
 > POTS connection...

And when there was variable cost, or when a cost-sharing system was used, this
was an inaccurate guess, not the actual cost in currency.  The values ended up
being disconnected from cost in currency and simply used as a weight.

So, at the end of it all, how is the cost field *currently used* on Fidonet
aside from "usually set to zero", and what sort of SHOULD clauses should be
added?

Saying it SHOULD be set modified before forwarding, or SHOULD be ignored on
import, or anything else seems to imply usage that doesn't really exist on
Fidonet today.
--- SBBSecho 2.32-FreeBSD
 * Origin: BBSDev.net - The BBS Developers Network (1:103/1)