cocoon-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Daniel Fagerstrom <dani...@nada.kth.se>
Subject Re: Supported and unsupported blocks
Date Wed, 16 Mar 2005 10:29:50 GMT
Torsten Curdt wrote:

>>>>...Where shall we draw the line between "supported" and
>>>>"unsupported"? Is it really the "two committers rule" that I applied
>>>>above?...
>>>>        
>>>>
>>>How about adding another condition: a block can only have "supported"
>>>status if we have automated tests for all its critical functions?
>>>      
>>>
>>That's a very interesting point.
>>+1, even if I'm not the most productive test writer :-)
>>    
>>
>>>This might make a big difference in the accountability of supported
>>>stuff in our releases.
>>>      
>>>
>>Yep. But this also means the core is currently unsupported ;-P
>>    
>>
>
>Hehe... Well, time to write some test then ;-P
>
>Actually I somehow like the idea of required tests!
>And it would also clearly draw the line. Much better
>than the two-committer rule.
>  
>
Don't agree at all ;) If I have to choose between an abandoned one man 
show with full junit test support (whatever that means) and a block with 
an active community but no tests there would need to be *very* strong 
other advantages for the one man show, for me to even consider it.

And take the >2 commiter rule as a starting point, if you want something 
to become supported, give us good reasons a start a vote.

>...but a lot of work :-/
>  
>
I agree that test driven programming is a an efficient and good way of 
developing code, and I have made good use of it in e.g. the JXTG 
refactoring. But as for everything else it is the quality of the tests 
that counts. It is far to easy to write tons of testing code that 
resembles the commenting style where you tell that "setFoo()" sets foo.

We should definively encourage our selves to write test code when we 
refactor things. It is useful for checking that we don't break anything 
and also documents our learnings about what the code does. Starting to 
write test code for allready existing code just because it "should be 
tested" is more than "a lot of work" it is IMO pretty close to wasting time.

/Daniel


Mime
View raw message