Re: ZNG dicussion

At 10:07 28-09-2001 +1000, Alan Kent wrote:

>   "I guess providing a 'web service' for access to information resources
>    to any person with MS Windows on their desktop in a standard way is
>    not that important to libraries."
>
>but I decided not to! :-) :-) :-)
>It would sound sarcastic when all I really wanted to say is I think
>a standard web service to existing Z39.50 resources would sharply
>increase the accessibility of library information to applications
>outside of a library. Whether such need exists, I do not know.
>This would not replace Z39.50. It would solve a different need.

It might be interesting to actually make a realistic list of such outside 
applications that might implement ZNG but who are not doing so at the 
moment. It could help nail down the actual requirements for ZNG, and lift 
it from the level of a geek activity ("let's do this because it's cool") 
into something that may address real problems... I don't have to implement 
ZNG to know that it would work or that it would be very simple to use for 
very simple applications. What I don't know is whether it has any concrete 
value *in the current landscape* of implementation and Z39.50 use. I think 
at least some of us who are concerned are not just reacting negatively to 
this because it is new, but because we have an interest in, or commitments 
in, some pretty complicated situations involving installed base, and some 
careful, semi-political games with vendors who have to invest time and 
energy in Z*.

As you play in this game, you come to realise that the actual technology 
plays a comparatively small role compared to the "political" maneuvers. 
John Kunze once put it very nicely when he told me that the real victory of 
Z39.50 was not really in its engineering or design, but in the consensus 
that it represented between a large and diverse implementor community. 
Having watched a few Z39.50-inspired "lightweight" search protocols rise 
(sort of) and fall over the years, I am less optimistic than some about the 
guaranteed success of ZNG.

But I'm also not too enthusiastic over some of the purely technological 
arguments.. yes, it is true that basic ZNG servers are trivially simple to 
write, and so are basic clients which only access one server and block 
until the result comes back. But if you need a more sophisticated 
application that queries a number of targets efficiently in parallel, then 
many of the most basic tools for implementing HTTP-like protocols break 
down, and you get back to doing some programming that, to abuse a motaphor, 
is not for people who enjoy eating quiche 
(http://www.cirr.com/~barkley/jokes/realprog.html)..

--Sebastian

Received on Friday, 28 September 2001 08:59:34 UTC