From: "Marc Cromme" <[log in to unmask]>
> I might not be on the main stream of how others feel about this, but I
> think it is no good idea to throw in a new error message because one
> implementor wants it.
I think that from our many years of developing diagnostics (Z39.50 and SRW)
we generally believe:
(1) we should avoid multiple diagnostics for the same condition; but
(2) we don't need to establish that a particular diagnostic meets any
"usefullness test"; there is no substantive harm in defining a diagnostic
for just one implementor.