From wpollack@DB.Stanford.EDU Thu Aug 27 08:19:43 1998
Date: Wed, 26 Aug 1998 17:41:31 -0700
From: Woody Pollack <wpollack@DB.Stanford.EDU>
To: chaims@DB.Stanford.EDU
Subject: minutes: 8/26/98

Date[ August 26, 1998]
Who[rb, lm, gs, wp]

Minutes[

Laurence talked about IBM's work-flow tool FlowMark and then we discussed a
bit how it compares with CHAIMS.  Decided we really had a couple of
advantages over that system conceptually:
	1) Estimate and using it to dynamically reconfigure our clients at
runtime
	2) partial data retrieval (i.e. not necessarily wait for a process
to complete and return all results, but being able to grab some of the
results in the middle)

Discussion about ClientIDs and where they should be generated.
	Ron pointed out that in order to maintain a stateless server
(megamodule), which is what we're aiming for, we would need the clientID to
be generated by the client/compiler.

Had a general discussion about stateless versus stateful (I doubt this is
what you call it) servers

Things to think about:
	* Possible second suite of megamodules to write as another demo of
chaims
	* As always, look for Ron's "Killer application" that can be done
trivially in chaims, but requires a bit of effort without chaims


]

Errors and suggestions are always welcome


woody

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Woody Pollack
wpollack@db.stanford.edu
http://www-db.stanford.edu/~wpollack
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~