[Asterisk-Users] *** Asterisk Sunday News: Off track with 1.0, moving forward

[Nicholas Bachmann]

TC wrote:

>>Does Asterisk have a test plan for releases?  It seems like if there was
>>a plan for testing that people could carry out (in a distributed
>>manner), our releases might not have so many quirky bugs.  Most testing
>>can be automated; only some would have to be done in an interactive
>>manner with real people.
>>
>>Nick
>>    
>>
>Is Nick volunteering to write test cases (for the next 12 mths)
>  
>
Yes, I would happily write some of them.

>I would love to see how we could write automated tests for all the different
>config combinations for all the different uses that lead to 'so many of the
>quirky bugs'
>  
>
Well, I'm not suggesting we immediately write a test case for every 
possible function.  However, the functions of major components like SIP 
and IAX don't change very much.  So, automated test cases to transfer 
calls, put calls on hold, etc. could be static from version to version.

>We might had a chance in hell if we wrote a test case before any functions were
>written
>  
>
Test cases for modules can be written by module authors (who know what 
to test for anyway).  Ideally, having test cases would be requisite for 
a module's inclusion in the main Asterisk distribution.

>back porting test cases to what is it 80K lines an growing is a project the
>size of asterisk
>  
>
What would we be back porting?  I'm talking about starting now.

>please tell talk up if you know some trick here :)
>  
>
Yes, start slowly and move methodically.  Software products twice 
Asterisk's size have test plans.  And anything methodical is better than 
haphazard the GIHTW* method currently in use.

Nick

*Gee, I hope this works!