Re: philosophy Re: Conformance requirements on browsers

On Mon, 16 Sep 2013 07:29:26 +0200, Charles McCathie Nevile  
<chaals@yandex-team.ru> wrote:

>> That can be a valid argument, but, if it is required for compat, its  
>> behavior should be in the spec.
>
> This assumes that the requirement for compatibility or the spec's  
> coverage
> of everything is sufficiently universal. That seems like a poor
> assumption. If it were true, I would be worried about whether the spec
> would be doing the right task.
>
> The web is *different* in different countries, languages and markets. The
> extent to which it should meet everyone's requirements is a question of
> judgement. But it would appear that we can judge the question of whether
> it *does* by the extent to which they are prepared to change things to
> meet it. People's perceptions of their market is unlikely to be perfect.  
> I
> believe that includes the Working Group and its editors.
>
> So the fact that a spec forbids something strikes me as a poor argument
> that it is therefore not needed in some market.

Yes. But that's not what I'm saying. I'm saying the opposite. If some  
vendor makes the case that something is needed for compat in some market,  
we should add it to the spec (assuming other vendors don't object). This  
happens from time to time (e.g. <keygen>).

>> In the case of <bgsound>, Gecko and WebKit never found it being big  
>> enough compat loss not to support it, and Presto even intentionally  
>> dropped support.
>
> Sure. I fail to see what useful information this provides about whether  
> it
> matters to IE.

What other browsers do is a useful data point when a browser vendor is  
deciding whether to support something or not. It's not the only data  
point, but one that is often used.

> More generally, it still appears that the assertion "things not permitted
> by the spec are forbidden" is basically untrue,

Not in general, I think. In general, the spec requires what to do, and  
what it doesn't say must not be done. In the case of supporting elements  
not defined in the spec, the spec allows it with a "should not".

> and that the assertion
> that supporting elements not defined in the HTML5 spec (or in WHAT-WG
> drafts as of today) makes a browser non-conformant is explicitly untrue.

Yes. But a UA claiming to be conforming would have to state what the  
reasons are for violating the "should not" in order to be convincing. :-)

> For what it's worth, that seems like the right situation to me.

-- 
Simon Pieters
Opera Software

Received on Monday, 16 September 2013 09:09:06 UTC