[TIP] Ideology

holger krekel holger at merlinux.eu
Tue Apr 28 04:27:05 PDT 2009

Hi Michael, 

On Tue, Apr 28, 2009 at 12:00 +0100, Michael Foord wrote:
> holger krekel wrote:
>> As to compasability: i think that nose and py.test both aim to provide 
>> their functionality through plugins these days.  This
>> sure enough comes with the cost of invoking plugin hooks but
>> ideally it should lead to the composability you wish for.    
> How wild an idea is it to provide *some* compatibility with the plugin  
> system for unittest? Are they just too different or could a subset work?

no clue currently.  I hope to look into your
3k/unittest.py work in the upcoming next two months. 
My guess is that most plugins are quite simple - so i 
guess we might want to pick some examples and see what 
it practically takes to implement a plugin for multiple
systems.   Are you up for a bit of sprinting around 
EuroPython? :)

>> On a sidenote, i guess if i started to read the source code of
>> the linux kernel and firefox, i'd find tons of code i do not
>> need for my usage - does this warrant to go for building my own OS and 
>> web browser?
> I use unittest because it does 'enough', and none of the extras provided  
> by nose or py.test are compelling enough to make me want to jump ship  
> (yet...).

fine enough :) 

> Also doing a lot of stuff with IronPython (which neither nose nor  
> py.test supports particularly well) it is nice to use the same test  
> framework for all my projects.

sure, very valid point. 

> The fact that they provide stuff I will never use (every large framework  
> has that) is not a reason why I don't use them.

right, although i think that complexity in debugging 
is a very valid concern - which is why test frameworks should
be particularly well tested on all kinds of levels. 


More information about the testing-in-python mailing list