-=> Quoting MIKE FARRILL to RICHARD TOWN <=-
 MF> I guess the courier has the 512 byte block (it better for that price
Don't think so
 -> pre-compressed files since an effect called "zero-stuffing" accounts
 -> for the slight decrease in apparent effective thruput noticed
 MF> How much does this zero stuffing affect cps?
50-100 or so cps (as measured on fully compressed files greater than 130kb)
Seems to vary as with the age of build of the BT-LZ algorithms used too
 
 -> If by better you mean better for maintaining a higher thruput on duff
 -> lines then, in the absence of LAP-M with SREJ, MNP10's good, since
 -> it'll go down to 8 byte resends.  But it does carry more overhead...
 MF> I thought that MNP10 was a cellular protocol.
That's MNP10 with ACE, which is also referred to as MNP10ECC
This dynamically links the line level to the error correction so that
any cellular flutter, or inperceptable to the ear phase differences
which can occur when changing cells, causes resends instead of the
line dropping.  Line level from modem to cellphone is critical so it
can be preset; also during startup negotiation of link can be forced
to be at a low speed, then as the link stabilises the connect speed
increases.
MNP10 on its own is just the error correction without the line level
dynamic link.  Its slow speed startup is either 1k2, 4k8, or max speed
the link will support
rgdZ
Richard
--- FMail/386 1.02
---------------