> Oh, OK, seems like I misread you.  I think what you said was "We spent
> years designing Z39.50 diagnostics so that clients would do something
> intelligent with them, before completely repudiating the idea", which
> seemed pretty strongly anti.

Actually, I was stating historical reality (not taking a position): the
Z39.50 community pretty much repudiated the idea, between the 1995 and 200x
version.  As I saw it: client implementors were saying we're not going to
develop clients smart enough to do anything useful with diagnostic
information; server implementors were saying we're not going to send any of
this diagnostic information anyway; so let's all stay with simple
diagnostics. That's how I saw it. I wasn't happy about it then. (The "we" in
my quote above didn't include me. Bad grammar.) If the SRW client and server
implementors feel differently then I'm happy.

>   So I think it's good to use solid,
> well-defined addInfo where appropriate,

Who could argue with that?