Far much valuable reply than mine :)
A great summary of common concerns that must be evaluated before starting coding a rule
system.
Also consider :
- using rule flow (if you are allowed to)
- use sequential mode and pipe your data 'in the next rule box'
programmatically (using rules?)
- ask your architect ? He is true for no-loop, but for others, re-writing existing DRL
code that used all that stuff, and throw all at once is farly more dangerous ...
> "remove all salience from rules and make the rules
independent (don't decide its execution flow)"
This is all the question.
It mainly depends of the usage, as Davide said.
It also depends of who is authoring/specifying rules : if the guy supposes some flow, he
is not writing independant rules ...
these other rules depend of the first :
2.no drools functions
3.Activation group / ruleflow / salience ...
5.don't call any java function (thus no functions at all, no predicates/utils
methods, only DSL? only POJO?).
this one is suspect ...
4.Don't use UPDATE and Global in drools .