Le Chaud Lapin <jaibuduvin@gmail.com> writes:

> Another thing to be wary of: there are people who insist on taking the
> concept of layering literally. When they go to implement an "OSI
> protocol stack", they will actually try to structure the software in
> layers, right up to Layer 7.

Very true. Even the OSI protocol development effort ran into this
snag. It took them a couple of years to realize they needed to turn
the upper layers on their side, with application "layer" using
both presentation and application services to , but with the
application itself being the one that dealt with the lower layers
directly, since "presentation" really has nothing to do with the
communication path.

> And finally: there are people who would swear that there is floating
> around somewhere a physical embodiment of OSI protocol stack.

I'm not sure what this means. There were several physical OSI protocol
implementations available (even if you don't include things like X400
and X500 that didn't follow the OSI model in any reasonable sense, but
were blessed as "OSI protocols" because the OSI community had
political needs for working protocols). There were even testing
companies that made sure the implementations conformed to the defined
protocols. No one used those protocol stacks for anything other than
checking boxes and, maybe a little, playing around, but I can assure
you they existed, for all they were worth.
-don