[rules-users] Existing Test Harnesses for Drools?

[Salina Fung/UFL - ICIL]

Dear Ansgar,

I am looking into the tools too.  Appreciate if you can share your
experience


Thanks
Salina


----- Original Message -----
From: Ansgar Konermann
[mailto:ansgar.konermann at googlemail.com]
To:
rules-users at lists.jboss.org
Sent: Wed, 13 Apr 2011 01:14:10 +0800
Subject:
Re: [rules-users] Existing Test Harnesses for Drools?


> On 12.04.2011 14:46, John Peterson wrote:
> > I was wondering if anyone out there is aware of any Test Harnesses
> > that could be used for regression testing Drools applications out there? 
> >  
> > We found this spreadsheet tester from Michael Neale:
> > _https://github.com/michaelneale/rule-spreadsheet-tester/#readme_
> >  
> > Are there any others?
> 
> Hi John, hi all,
> 
> we used a similar approach (Excel) for some time, but dropped it well
> over a year ago because it's very hard to maintain. We're now using
> plain Java tests to load *.drl files and run a plethora of tiny unit
> tests against individual rules. A few larger tests load most of our
> rules and execute integration test scenarios.
> 
> Regarding maintainability: we're using a pojo based fact model to
> represent our knowledge which needs to be changed from time to time,
> driven by new features. This is easier if tests are in Java than if they
> were Excel. Excel cannot be diffed or merged easily, as it is a binary
> format. Besides, you normally cannot use the search function(s) provided
> by modern IDEs to search inside Excel files. Plus you have code
> completion to code your test setup and the assertions if using
> Java-based tests.
> 
> We do have our own (internal) test harness, which is basically a set of
> reusable test superclasses and helpers. I heard rumors from our legal
> dept. that it would be okay to open-source it one day, but due to lack
> of time nobody ever really did.
> 
> If you want to follow this road, let me know. Maybe I can provide some
> more insights and/or code examples.
> 
> That said, I think Excel-based tests might be okay in scenarios where
> the underlying fact model does not change frequently. For us, they have
> proven too heavy to maintain.
> 
> Best regards
> 
> Ansgar
> 
> >  
> > Thanks!
> >  
> >
> >
> > _______________________________________________
> > rules-users mailing list
> > rules-users at lists.jboss.org
> > https://lists.jboss.org/mailman/listinfo/rules-users
> 
>