Matthew Toseland wrote:
> On Tue, Mar 20, 2007 at 06:06:01PM +0100, Sback wrote:
>
>> Florent Daigni?re (NextGen$) wrote:
>>
>>> * Sback <sback at sback.it> [2007-03-20 15:44:43]:
>>>
>>>
>>>> Deliverables:
>>>> * Create specific method tests with a fine granularity on most used code
>>>>
>>> I'm not sure we want tests to focus on the most used code : if it breaks
>>> we will realize it soon enough...
>>>
>
> My concern here is that he may write a whole load of tests and then
> we'll go and rewrite that section of code (e.g. when rewriting the
> congestion control/retransmission/link crypto code) and the tests won't
> compile any more, or will test things that we no longer use.
>
To avoid this problem, perhaps I could generate a detailed report
about what parts I believe should be tested. Then I can submit it
to the developers to your suggestion.
Using a profiler to view the most used code was an idea, but
of course it won't be useful doing useless things :s
We have the same target: good tests!
Of course after some big modification in freenet class behaviour tests
have to be changed. But I hope, apart from GSoC, to continue
a contribution on Freenet and I would be happy to keep them updated
as long as they are useful.
And writing tests I could understand better the underlaying code and
maybe offer my contribution in developing.
>> You are right.
>>
>>
[cut]
