[TIP] structure of a testing talk

[Jonathan Hartley]

On 02/10/2012 13:24, andrea crotti wrote:
> 2012/10/2 andrea crotti <andrea.crotti.0 at gmail.com>:
>> 2012/10/2 Ned Batchelder <ned at nedbatchelder.com>:
>>>
>>> Andrea, this is helpful to me.  I've proposed a Getting Started Testing
>>> tutorial for PyCon, and it's great to see how other people tell the story.
>>> My tutorial is an expansion of a talk I did last year at Boston Python.  You
>>> can see my slides from that talk at
>>> http://nedbatchelder.com/text/starttest.html .  I'll accept tweaks to my
>>> story line too!
>>>
>>> --Ned.
> Another couple of things which I think are useful and I didn't see are:
> - pure functions and side effects
>
> - why in Python it's even more important than other languages to test
>    (I'm speaking to C programmers mainly here, so might depend on the
>    audience). And maybe some examples of how things can go horribly
>    wrong only at run-time even if there was a clear big mistake in the
>    code.
>
> Apart from that I might steal many ideas, but I only have until Friday to
> prepare the talk, so I won't be able to explain everything..
>
> Another thing which I will do is to use IPython+Emacs a lot, to show
> actual live code.
>
> And about this a nice little project from PyconUK might also be
> useful.  This allows you to replay code that you already wrote in an
> interpreter, in a presentation style, avoid typing but keeps the talk
> interactive at the same time:
>
> https://github.com/inglesp/prescons
>
> _______________________________________________
> testing-in-python mailing list
> testing-in-python at lists.idyll.org
> http://lists.idyll.org/listinfo/testing-in-python
>

I think I disagree with the idea that testing is more important in 
Python than it is in other (presumably statically typed) languages. I 
think it's equally crucial in both. Am I deceiving myself?

I understand that Python experiences runtime errors which some other 
languages would have caught at compile time.

But I don't think this means there are any 'extra' tests that one has to 
write in Python which one wouldn't write for other languages. The tests 
don't have to be more strict, nor are they any harder to write (the 
opposite!) If all tests pass, the program can be deployed, otherwise, it 
cannot.

The fact that some test failures cause a runtime error in Python versus 
a compile time error in C doesn't seem to have any practical impact on 
this situation.

Am I'm overlooking something? If so, I'd love to be educated.

I confess: I'm thinking in terms of old-fashioned 'static typing' such 
as C, C++, C#, Java. I'm not familiar enough with a modern Haskell-style 
type system to comment on how that affects things.

Best regards.

     Jonathan

-- 
Jonathan Hartley    tartley at tartley.com    http://tartley.com
Made of meat.       +44 7737 062 225       twitter/skype: tartley