Re: Microsoft's cross-document messaging and DOM Storage tests published

On Thu, 29 Jan 2009 01:38:44 +0100, Ian Hickson <ian@hixie.ch> wrote:

> On Wed, 28 Jan 2009, Dan Connolly wrote:
>>
>> Since lots of developers will use the tests without reading the spec, I
>> prefer to have both the spec and the tests be normative. If they
>> conflict, there's a bug, and I don't see much value in pre-judging where
>> the bug is.
>
> The reason not have the tests normative is that they can often easily
> implicitly require behavior that is not intended to be required, without
> it being at all obvious. (In a spec it's usually obvious when that
> happens, since there has to be a "must" somewhere.)
>
> Also, it's much easier for tests to implicitly require a particular model
> without that model being thought through than it is for a spec to do  
> that, since in the spec anything not said explicitly isn't required.

Indeed. While in practical terms many people will use tests and not  
completely understand the spec, I think it is valuable to name one thing  
that is meant to be a priori right in cases of conflict.

If there is a difference, and ergo a bug, it is true that both the spec  
and the test should be looked at to see the best way of making the world  
clearer and better, but allowing multiple normative definitions doesn't  
seem terribly helpful. Making one source be normative ensures that if we  
have a working process, we know where to look - and if we don't have a  
working process then multiple sources won't save us.

So I think the tests should be agreed by the WG one by one, but I don't  
think they should be normative.

cheers

Chaals

-- 
Charles McCathieNevile  Opera Software, Standards Group
     je parle français -- hablo español -- jeg lærer norsk
http://my.opera.com/chaals       Try Opera: http://www.opera.com

Received on Thursday, 29 January 2009 07:18:02 UTC