🚀 go-pugleaf

Several big problems with MicroCom MNP were not mentioned in the
interview.  First, while there are N companies who have implemented the
protocol, there is no immediate way of knowing if the implementations
are tuned.  MicroCom does have seminars in implemting MNP, but companies
are often interested in reienventing (or perhaps reiemplemting) the
wheel.  In fact, since MicroCom sells modems itself, it has a vested
interest in having its implementation of MNP work "better" than anyone
else's.

Only Class 1 through Class 3 MNP are free for other modem companies to
use.  Class 3 only has about a 10% increase in efficiency over the
carrier speed.  The dramatic throughput increases start appearing in
Class 4 and Class 5, which must be licensed.  I'd be interested in
knowing what percentage of non-MicroCom MNP modems are above and below
the Class 3 boundary.

There are already modem manufacturers that are using a "superset" of
MNP.  Some manufacturers are extending capabilities of MNP; some are
circumventing the royalties associated with the higher-level
implementations.  These actions will limit the universality of the
high-performance MNP levels.

Finally, as has been noted in Telecom already, no modem-to-modem
packetization protocol can guarantee reliable end-to-end transmission of
information.  There must be some additional mechanism to assure that the
information is correctly getting transferred.  Any reliable byte stream
must have its packetization/retransmission facilities exist in the
source and destination machines.  By its implementation, MNP is not be
the answer for error-free communications.