[alicebot-archcomm] Why not just define a NTP service?

[Gary Robertson]

Hi Ernest,

I disagree with your assertion that  my suggested protocol is cumbersome 
for what it buys the developer who uses it.  It provides for detailed help 
documents, it provides for defaults for common usage, and it provides for 
error help when it is encountered.  But the best part about it is that it 
provides for one set of reusable interface standards that could leverage an 
aiml developer's access to dozens and maybe eventually hundreds of valuable 
services.  One aspect I like best about this approach is that its use is 
"immediate" and LANGUAGE INDEPENDENT from the aiml developer's perspective.

I chose to use getTime because we generally perceive it to be a 
simple-stupid service, but it is not.  Already you have directed me to two 
distinct services to handle the features I intend to handle in one service, 
and neither handles time zone selection or a daylight savings time switch 
(not everyone uses daylight savings time).

I do not propose that this be an exclusive arrangement by any means.  But 
using one set of interface rules with an intermediary server, with up to 
two backup servers as I have suggested,  has got to be a lot better than 
each aiml developer having to learn how to interface with every service 
independently with unique ways of handling each one.

I can see a number of independent aimlService Servers sprouting up to 
answer the call to meet the needs of the aiml community.

In any event my goal is to get a working protocol between AIML and my 
machine that is simple, straight forward, and robust enough to get the job 
done.  If you develop one that manages your view of the world but can also 
work for mine, I am a happier man.  I now have a test platform ready and 
waiting to try yours out.

Bring it on!

-gr