launchpad-dev team mailing list archive

[Deryck Hodge <deryck.hodge@xxxxxxxxxxxxx>]

On Tue, Nov 2, 2010 at 9:18 PM, Deryck Hodge <deryck.hodge@xxxxxxxxxxxxx> wrote:
> I don't think the Windmill tests are integration or end to end tests.
> That's what they aspire to be. :)  They are at best interaction tests.
>  Also, I share you perspective in terms of the ideal for testing, but
> Windmill does not provide this ideal.  I do agree we should test AJAX
> and JS interactions, though, which is why I'm in favor of leaving
> Windmill to run yui unit tests.  If we came up with a clear story for
> how to do reliable, useful, and fast end to end testing, I would of
> course support that, but I don't think Windmill can ever be that.
> That's a different issue, IMHO, then asking if the tests as they are
> now have any value.
>

To clarify what I meant, a bit from a chat Francis and I are having:

<flacoste> why aren't they integration tests?
<deryck> because they focus on very minor single interactions.  I
think of them as interaction tests.  It's arguing over semantics,
maybe.  But I would see the combination of what we do in page tests
and windmill as closer to true integration tests.
<deryck> and people avoid writing windmill and structure the tests and
dev work to test the least as possible in windmill, so it's hard for
me to think of them in true integration terms for that reason
<deryck> integration tests imply a complete story to me.... user
starts at X ends up at Z and Y happens -- across all the site, not
just for comments or branch linking.

Cheers,
deryck



-- 
Deryck Hodge
https://launchpad.net/~deryck
http://www.devurandom.org/