Re: [xml-dev] Testing XML don't use xUnit

[Andrew Welch <andrew.j.welch@gmail.com>]

Yay, more...

> Absolutely true. Devs are kings. That doesn't mean that everything
> they are given to do will be done competently. It means that
> everything that they are given to do will be done to the limit of
> their competence and sometimes their competence and the techniques
> they are used to doesn't stretch that far. Like for example here
> http://ravimohan.blogspot.co.uk/2007/04/learning-from-sudoku-solvers.html

Sorry that's a lot to read - what's the tldr; summary?

It does mention xslt and sudoku solvers, so I won't pass up the
opportunity of posting the link (again :) to my own:

http://andrewjwelch.com/code/xslt/sudoku/sudoku-solver.html

: )


> Some are so cocky they think everything is autodidactable.

I don't know what that means, nor does google apparently.


>> Developers 'private testing'? There's nothing private about the tests
>> you deliver with the code.
>
> Private testing means whatever the developer did to verify his
> program. It doesn't matter whether he publishes it or not.


What sort of verifying and publishing do you mean?

Agile dev work requires that you write testable code and the tests
that go with it - it all must be delivered into the build.  The build
will repeatedly run those tests, as will all the other devs before
they deliver code.


>> Regarding regressions, Agile is all about
>> avoiding regressions and confident refactoring, all based on the tests
>> written by the developers.
>>
>
> which won't run if they are not designed to be robust enough to be
> repeatable and are unlikely to be repeatable unless they are planned
> and if they are planned why on earth wouldn't you document that.

By definition they have to be repeatable because they are run as part
of the continuous build repeatedly thousands of times.


>>  Read up on continuous build.  The codebase
>> is continuously tested - there is no 'regression pack' any more, its
>> all done as part of the build.
>>
>
> You're getting carried away with buzzwords.
>  http://sqa.stackexchange.com/questions/643/fitting-regression-testing-in-a-agile-scrum-development-cycle

What is that link telling me?   (continuous build is no more of a
buzzword than 'xml' these days)


>> Despite the numerous variations in how people 'do agile', that area
>> has been consistent since the beginning.
>>
>
> No it hasn't but never mind.

Sigh, it has.  The Martin Fowler article is good read on this:
http://martinfowler.com/articles/continuousIntegration.html


> There is a very simple reason for the success of TDD. It is not an
> inherently superior development methodology but it has one crucial
> feature that no other methodology has. It's the only methodology that
> has managed to trick developers into doing testing

!  That made me smile.

<this_will_sound_patronising>You should get some experience of Agile
and modern dev work in general, before continuing this thread</...

-- 
Andrew Welch
http://andrewjwelch.com