By "(Java)" do you mean Java code between calls to fireAllRules() or Java
code
in RHS? I think that what you describe in items 5, 6 and 7 can very well be
written as rules, based on a Validation object inserted at the time the
actual
operation is done in WM. It should contain:
- links to the old, or new, or both, Fact objects
- the validation status (OPEN, VALID, INVALID)
Need a more detailed outline?
-W
On 12 August 2010 10:08, Stephen Mcgruer <s0840449(a)sms.ed.ac.uk> wrote:
Hi there,
My project uses Drools to (among other things) validate new
facts/updates to old facts, according to a set of rules. (For example,
we might have a rule that says "No two people can be named the same
thing".) Our method of validation and error reporting was originally
chosen for a stateless session (before we had the requirement of doing
checks like the above), and I feel I've had to hack it a bit much to
get it working with stateful sessions. The current methodology is:
1. (Website) The user inserts/updates a fact.
2. (Java) If the fact is new, an object is created for it. If it is an
update, the pre-existing object is fetched from the object repository
(the interface to the database) and a backup copy of the object is
taken.
3. (Java) This fact is either inserted into the KnowledgeBase (KB) or
a pre-existing fact in the KB is updated depending.
4. (Rules) If the new/updated object fails a validation rule, it will
be retracted and a ValidationError object will be inserted into the KB.
5. (Java) The code then searches for ValidationError objects that
match the just inserted/updated object. If it finds any it records
them and then retracts them from the KB.
6. (Java) If there are no errors, then that's it - a success report is
generated and passed back to the website.
7. (Java) If there were errors, then if the user was updating a fact
the backup of it (before the changes) is inserted back into the KB.
The website is then passed a failure report.
It just feels very hacky to do so much manual insertion/retracting
from the Java side of the code.
So, is there a 'correct' way to do stateful validation? And does our
current way even work, or are there hidden problems that we've not
spotted?
Cheers,
Stephen
--
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.
_______________________________________________
rules-users mailing list
rules-users(a)lists.jboss.org
https://lists.jboss.org/mailman/listinfo/rules-users