On Mon, 22 Oct 2012, Brian E Carpenter wrote:
> On 18/10/2012 02:25, Noah Mendelsohn wrote:
>> On 10/17/2012 7:57 PM, Ian Hickson wrote:
>>> Yeah. Turns out we (the Web standards community) haven't been doing
>>> such a great job of making our specificatiosn match reality.:-(
>>
>> Um, true... but it's also the case that the implementation community
>> hasn't over the years been doing as much as might be best to make
>> reality match the specifications. The new specs we're writing now
>> would like have been a lot thinner and cleaner if they had.
>
> However, I think the concern here is that if certain IETF specs need to
> be updated to match the real world, that work needs to be done in the
> IETF.
That's certainly an important concern, but it's not the one I had in mind
when I made my comment.
Rather, while Ian is right that our new specifications must match today's
"reality", we should also be sobered by the price we're paying for years of
implementations that do things like sniffing. Yes, given that it's become
widespread, documenting ways to do it interoperably and securely is good.
I'm not convinced that implementing sniffing in widely-used user agents was
the right choice way back when.
So, let's try harder than we have in the past to not introduce yet more
such cruftiness, even if we do have to document what's there. Maybe we
should have been a little more hesitant to do things like having both SVG
and Canvas as ways of doing graphics. Perhaps that overlapping function is
ultimately justified, but there's a long term cost. We're moving too far
away from KISS IMO.
HTML5 is great work, but the spec is very, very thick. Keeping things as
small and simple as we can moving forward is really what I had in mind.
Noah