PE>> I've taken out the "-1" from zip (to make it compress
fast).  Now
 PE>> it's in the middle.  When transmission rates drop, as they sometimes
 PE>> do, perhaps because the MNP4 bug is raising its head, throughput
 PE>> drops, so the less the better.

 RS> Dunno. The MNP4 business isnt so much a bug as just a bit slower than
 RS> average. I normally get a 1550 thruput, as opposed to say 1650, so its
 RS> no big deal. My superficial reaction is that the speedup in the packing
 RS> time is worthwhile. But PKZip varys quite a bit in how the speed of
 RS> packing changes with the type of data being packed and I havent done any
 RS> tests on messages.

No, there is an MNP4 bug as well, which puts my modem into double-send when
connected to a Maestro.  Goes away if I pause the computer.  I'll see if I
can do some more tests re zip -1, zip -9 and zip based on the packet you
actually receive (have to get it while its happening). Naturally if you got
your mail via the point, it would be zip -9 all the way!

Another thing re connects.  Seems that locking your rate at 14400 hasn't
helped at all after all.  I am wondering whether *E5 or *E7 would help at
all from my end.  *E5 is Auto-reliable V42 with phase detection.  *E7 is
auto-reliable V42 without phase detection.  Currently I use *E9 which is
auto-reliable V42bis with phase detection.  Since you monitor your call,
wouldn't it be better to use V42 on your end and if the thing pulses, pause
for a few seconds?  With the amount of data you are getting now, I'm
wondering whether that would give an increase CPS rate, even with the
manual pause.  Maybe the MNP4 connect problems would go away too, which is
a further 30 seconds saving for each connect.  That is because my end
always gets an MNP4 connect, whilst yours doesn't always. Maybe we won't
see the same thing happening with V42.  BFN.

Paul

--- GoldED 2.40