[comp.binaries.ibm.pc.d] More on Kermit file transfers.

JRD@cc.usu.edu (Joe Doupnik) (11/19/89)

        Perhaps I could add a small word on the common subject of slowness of
Kermit file transfers. The first is the protocol was designed to work between
dissimilar hosts, through their frontends or other channels. These situations
assume that most control characters do not survive the trip, that only
printable characters stand a chance, and that the file systems at each end are
probably vastly different. Robustness under these trying conditions requires
mechanisms built into the protocol. Second, my personal experience has
indicated that transfer speed is frequently dominated by the skill of the
programmer rather than the inherent protocol. Third, the status display of
programs does consume a fair fraction of the transfer time. The simpler or
more hardware dependent it is then the faster things move.
        Btw, in the forthcoming release of MS Kermit the terminal emulator
is up to the DEC VT320++ level, and file transfers can be done with sliding
windows (even long packets in the windows).
        Joe D.