Thanks for the link, John.  This is a very good resource, and I must
have missed the link to it on the NISO site, if there is one.

>It explains that you don't implement NCIP as such, in the sense of
>implementing all 45+ services, let alone all the optional elements
>within them.

Isn't this just the problem, though.  If there was a decent C library
out there that 'did' NCIP, would it have been folded into a number of
different applications by now - perhaps even new uses that neither You
nor I had ever thought of.

>The expectation was, and still is, that implementations conform to
>"Application Profiles"

I see, but once again if different developers go off and develop
different 'bits' of the entire NCIP protocol, then there are no
advantages of scale.  Perhaps I am now revealing my relative ignorance
of NCIP but if a common project were to evolve in, say, Java wouldn't
the creation of abstract classes from which code for all the profiles
could be generated save duplication of effort.  Even in the case of a
more proceedural language, like C, there would be substantial scope for
code re-use.

Anyway, my questions don't come from some place of lofty understanding.
 I am rather ignorant of the low-level aspects of NCIP and I do expect
that such a project would require a lot of work.  Could it be made to
stick in a round of European Union funding, I wonder?

Best regards,

David Kane
WIT Libraries
http://library.wit.ie/
++353.51302838