That was the plan, I was thinking what work is coming out of the
ruleflow work can be turned into generic language features - detached
agenda groups was another idea - where rules can no longer be added
once it starts firing. Or another one is to clone facts once the
AgendaGroup is detached, so that even if the original changes, the
rule still fires with the original data.
Mark
Michael Neale wrote:
I think its a useful idea worth further exploration.
Of course, one needs to have a multi core dev machine to start properly ;)
Or hyperthreading at least (is that still around??).
Perhaps with rule flow, the "parallel potential" can be made
declaraively, by specifying that certain modules are concurrent in
mode, versus ones that cannot be.
On 12/27/06, *Mark Proctor* <[EMAIL PROTECTED]
<mailto:[EMAIL PROTECTED]>> wrote:
In drools there is no longer a single agenda, there are AgendaGroups
with MAIN acting as the default. I've been thinking that it
might be
useful to have named Threads too - so you can specify a thread
for the
AgendaGroup to execute in. This would allow an AgendaGroup to execute
its rules in parallel with other, and thus the MAIN, thread - however
fact assert, retract and modify will remain a single thread
for now -
so there could be race conditions between different threads fighting
for the next working memory action. But this can still be useful, I
think, especially for cases where you have rules with no
side effects.
Mark
---------------------------------------------------------------------
To unsubscribe from this list please visit:
http://xircles.codehaus.org/manage_email
<http://xircles.codehaus.org/manage_email>
