Hello,
For performance reasons, assuming I am using STREAM mode, is it better to :
A. create several streams in my rule file and inject objects into
various streams depending on my needs
or
B. create a different session for each stream that I might have had in
A. above?
And if I do use option A, is my understanding correct that each stream will
be processed on it's own thread?
And if so, I notice that I am able to update a common statistical object
that I created within my rule file and update it from consequences on
different streams. If each stream is processed on it's own thread, is
updating of this in rule defined statistics object synchronized/thread safe?
For example in my rule file I have the following "stats" object:
# stats object
declare ObjectStats
id : String @key
name: String
lastReadingA : int
lastReadingB : int
end
..then I have the following two rules..
rule "update stats from objects A"
lock-on-active
when
$stats : ObjectStats( $id : id)
$a : ObjectTypeA(parentId == $id, $aValue : value) from entry-point
"stream for objects of type a"
then
modify($stats) { lastReadingA = $aValue };
end
rule "update stats from objects B"
lock-on-active
when
$stats : ObjectStats( $id : id)
$b : ObjectTypeB(parentId == $id, $bValue : value) from entry-point
"stream for objects of type b"
then
modify($stats) { lastReadingB= $bValue };
end
This is working fine and the lock-on-active seems to be preventing me from
getting the "cross product" problem, but is each of these streams processed
independantly on their own thread?
Thanks,
Chris