W3C home > Mailing lists > Public > public-sws-ig@w3.org > September 2004

Re: granularity/definition of a "service"

From: David Martin <martin@AI.SRI.COM>
Date: Sun, 19 Sep 2004 17:05:46 -0700
Message-ID: <414E1EDA.5010907@ai.sri.com>
To: "Huhns, Michael" <huhns@engr.sc.edu>
CC: jeff@inf.ed.ac.uk, public-sws-ig@w3.org

Huhns, Michael wrote:

> Thanks! /dev/null is a good example, and I withdraw my claim that a
> receive-only service would not be useful.
> The "display" service you describe might have no output, but it does
> have an effect, so it seems more than a simple receive.

Well, depending on one's beliefs, one might well say the same about the 
"prayer service" that you suggested, Mike :-).

I like your examples, BTW.  I'm only responding because you asked me (a 
few messages back) what I had in mind.

What I had in mind were services that only ever involved a single 
message - such as the display service.  I was focusing on that aspect, 
and ignoring the question of whether they had any effects.

Another example in this vein would be an informal "survey", like when a 
TV station asks viewers to call or send email or visit a Web site to 
indicate "yes" or "no" in response to some question.  If the TV station 
sets up the receiving functionality as a Web service, that could be seen 
as a valid example of a receive-only service. From the point of view of 
an agent that is concerned with generating one of those messages, 
there's nothing else for it to deal with, besides the single input 
message.  That's why one could think of that as a receive-only service.

But I also consider these examples as exceptions; as I said before, in 
most cases I think it is *not* appropriate to think of a receive 
operation by itself as a service.


> -----Original Message-----
> From: jeff@inf.ed.ac.uk [mailto:jeff@inf.ed.ac.uk] 
> Sent: Sunday, September 19, 2004 12:42 PM
> To: Huhns, Michael
> Cc: David Martin; Monika Solanki; public-sws-ig@w3.org
> Subject: RE: granularity/definition of a "service"
> Quoting "Huhns, Michael" <huhns@engr.sc.edu>:
>>A "service" that only receives is equivalent to a write-only memory.
> I
>>have never found that to be a useful service and would like to hear
>>about the situation you are imagining where it would be a coherent
>>stand-alone functionality.
> /dev/null on Unix systems is often used as a sink to throw output
> away.
> Also, something might be receive-only as a service (ie, so far as
> it's description in OWL-S or whatever as a service was concerned)
> but nonetheless allow the data to be accessed in some other way.
> (Perhaps something that projected pictures on a screen would be
> an example.)
> -- Jeff
Received on Monday, 20 September 2004 00:06:18 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 17:32:46 UTC