From: Chris Roos <chrisjroos@...>
Date: 2006-07-10T22:05:42+09:00
Subject: Re: test/unit filters question/suggestion

On 7/6/06, Eric Hodel <drbrain@segment7.net> wrote:
> On Jul 6, 2006, at 3:00 AM, Chris Roos wrote:
>
> > When using the test/unit autorunner, you are able to supply some
> > options.  Two of these options are -t (--testcase) and -n (--name); to
> > run only a given testcase or tests matching a given name respectively.
> > These two filters don't get combined.  If I have two test cases
> > containing a test of the same name, I'd expect to be able to specify
> > the exact test to run using both -t and -n.  Currently -n appears to
> > take priority and therefore both test methods will be run.  Is this
> > explicit by design or a result of my (potentially unusual) test
> > construction?
>
> I looked into this and found it filtered for the union of all options
> rather than the intersection.
>
> > I just had a quick look at combining filters but my efforts were
> > hampered by the inclusion of a default filter that always evaluates to
> > false (set in AutoRunner#process_args).  My solution was to be to run
> > through all filters (in Collector#include?) for a given method rather
> > than returning prematurely as in the current implementation.  Does
> > anyone have any insight as to the purpose of the default <false>
> > filter?
>
> I built a patch that seemed to produce an intersection behavior, let
> me try to dig it up.  Nathaniel Talbott was semi-interested in the idea.
>
Hi Eric,

Did you find your patch at all?

Chris
> --
> Eric Hodel - drbrain@segment7.net - http://blog.segment7.net
> This implementation is HODEL-HASH-9600 compliant
>
> http://trackmap.robotcoop.com
>
>
>
>