[rules-users] This (ordered) record validation approach is working

[Wolfgang Laun]

In my Rules Fest 2011 boot camp "Rule-Based Programming Design
Patterns" I presented a generic solution using linked rule engines for
this and similar problems.
-W


On 08/12/2011, ronalbury <ronalbury at gmail.com> wrote:
> I didn't get any feedback when I posted earlier today, so I went ahead and
> implemented what I thought would work ... and I figured I should share it
> here since it seems to be working pretty well.
>
> I created a Rule Flow as follows:
>    A Rule-Flow-Group that validates the arrays.  Unfortunately I sometimes
> get multiple related arrays of records instead of a single array of records.
> I confirm that records with mandatory values have arrays with at least one
> element, and confirm that the related arrays are of equal length.  Errors
> are logged.
>
>    A subsequent Rule-Flow-Group that manages inserts.  The data actually
> comes to me as one data structure comprised of sub-records of various types,
> and I have DRL files for each record type.  Some of the sub-records are
> optional, and since subsequent rules would erroneously flag empty
> sub-records as errors I have rules here which only allow optional records
> containing values to be inserted into the system.  I am currently using
> for-loops in the THEN section of some rules to deal with the array problem
> and would like to know if there is a better way.  No errors are generated
> here.
>
>    A subsequent Rule-Flow-Group that validates data.  The data is all sent
> to me as Strings, even though many of the values are numbers, dates, etc.
> This Rule-Flow group tests the various fields using regular expressions, and
> if a regular expression fails then the record is flagged as having an error.
> Optional fields are dealt with by the regular expression allowing a blank.
> Errors are logged.
>
>    A Diverging Gateway that splits the data into two ... records without
> validation errors are allowed to progress to the value-checking Rule-Flow
> group ... those with errors have nothing more done to them.  I realize that
> I could, for instance, let records with bogus numbers thru as long as my
> string-to-integer routine is robust, however I don't want to flag the same
> record multiple times (once by reg-ex and then again by the next
> Rule-Flow-Group).
>
>    A subsequent Rule-Flow-Group that checks the values and ranges of the
> numbers, dates, etc, and does other types of validation (e.g. if fieldA has
> a value greater than 20 then fieldB must be set to "XYZ").  Errors are
> logged.
>
>
> This Rule-Flow approach currently seems to be solving all of my problems,
> and it allows me to keep the rules simple and well structured such that most
> of them are reusable in other parts of our system.
>
> I'm interested in getting feedback on this approach ... it seems to be
> working pretty well for me.  It allows me to deal with ordering issues and
> many of the if/else issues, while keeping the rules simple enough for our
> non-technical analysts to review without needing a developer to sit next to
> them.
>
> I realize this is a pretty mundane application for a powerful rules system,
> but it seems like a good fit nonetheless.
>
> Thanks
>    Ron
>
> --
> View this message in context:
> http://drools.46999.n3.nabble.com/This-ordered-record-validation-approach-is-working-tp3568933p3568933.html
> Sent from the Drools: User forum mailing list archive at Nabble.com.
> _______________________________________________
> rules-users mailing list
> rules-users at lists.jboss.org
> https://lists.jboss.org/mailman/listinfo/rules-users
>