Re: [widgets] Test suite questions

On Wed, Sep 30, 2009 at 10:56 AM, Robin Berjon <robin@berjon.com> wrote:
> On Sep 29, 2009, at 18:14 , Marcos Caceres wrote:
>>
>> ----
>> ta-VngNBkhUXz:
>>        "If the protocol used for acquisition of a potential Zip archive
>> does
>> not provide, or otherwise include, a media type, then a user agent
>> should treat the acquired potential Zip archive as if it has been
>> acquired from a protocol that does not label resources with a media
>> type."
>>
>> I agree that "should" is evil. But I think here it is justified.
>
> Well, it might be more an issue with phrasing. What the above is saying is
> essentially: IF protocol does not label zip with media type THEN look at zip
> as if coming from a protocol that does not label with media type. If green,
> treat as if it were green.

No: it says if is green, then the UA SHOULD treat as green.

Unsure how to proceed.

>> ----
>> ta-HTgovPjElK:
>>        "During Step 7, a user agent must apply the algorithm to process a
>> configuration document."
>>
>> I don't think this assertions is redundant. It can be tested by having
>> a configuration document that exhaustively tests a UA by having a
>> complex configuration document that covers all test conditions. This
>> tests would be "compound" (as opposed to atomic), but I think it's an
>> important test.
>
> We discussed this and I don't think that it's possible to craft a document
> that tests all of the branches of the algorithm at once. This means one has
> to devolve to multiple tests. Those might be somewhat more complex than the
> more unit-oriented tests that we have, but not really by that much.

Agreed.



-- 
Marcos Caceres
http://datadriven.com.au

Received on Tuesday, 20 October 2009 14:09:43 UTC