These are precisely the vagaries that prompted me to ask the question.

In some ways it would seem, IMNSHO, that some are confusing inter-operability with open.

To this day, I haven't found an universally agreed upon definition of open.  Usually it means whatever IBM wants it to mean.

The are so many "layers" to a computer system that a broadly accepted definition is difficult at best.

At the software level: does open mean that I can take my binaries from machine A, copy them to machine B and expect them to run the same way as they did on machine A?  If this is so, POSIX has been a great leap in this direction.

At the hardware level: does open mean that I can take any peripheral or device from machine A, connect it to machine B and have it work the same way as it did on machine A?  If this is so, then PC's are open (thanks to ISA).

At the network level: does open mean that machine A plays nicely with machine B?  If so then HP, Unisys, DG and many others have been open for decades (had to be to work in an IBM dominated world).

I'm sure that I have left out many other permutations of comparisons, but I think it is enough to get a feel to the depth of the situation.

Personally, "open" is a marketing gimmick.  IT is responsible for actually verifying that what is used will do what is required.  No amount of "open" hype substitutes for fulfilling a fiduciary responsibility.

>>> James Clark <[log in to unmask]> 08/25/99 10:22AM >>>
Isn't the term OPEN have to do with the ability to accomplish work on a
computer. POSIX was just and is a API definition and expectation of how
<snip>