Implementing the Test-Run vision directly above T::P/TAP::Harness

[Andy Armstrong]

On 2 Oct 2007, at 20:50, Shlomi Fish wrote:
> Some of the enhancements I had in mind (or already implemented) are:
>
> 1. Colours - for the summary line and for the individual file  
> verdicts.
>
> 2. Running the test scripts using alternate, customisable  
> interpreters.
>
> 3. Customise the output using templates, Text::Sprintf::Named, etc.  
> (to
> eliminate the final percents, etc.)
>
> 4. Collecting the statistics and data for the entire run, and  
> preparing XML,
> HTML, etc. reports based on it.
>
> 5. Performing tests over the wire using TCP/IP, HTTP, etc.

Yes please! 3 and 5 are particularly topical - but those are all  
things we're interested in.

> Some of the things I'm looking at for a potential codebase are:
>
> 1. Short methods for easy inheritance and over-riding. That or at  
> least the
> willingness to allow me to factor the code appropriately. (Long  
> methods are
> in any case, identified as a bad code "smell" in Martin Fowler's  
> Refactoring
> book)

We have long methods - and excuses to go with them :)

> 2. If there are helper objects which I may wish to over-ride, then  
> I need the
> ability to specify roles or plugins for them.

Is this the Test::Harness 2.99 codebase we're talking about? Sometime  
soon we need to write a cookbook explaining how our various  
interfaces are used. This sounds like a good opportunity to exercise  
them and learn how they look from an external perspective.

-- 
Andy Armstrong, Hexten