Re: Removing optional constraints (Re: Overconstrained)

On 26 March 2013 04:08, Harald Alvestrand <harald@alvestrand.no> wrote:
> Changing the subject line again, since "overconstrained" is another
> topic....

You just don't like my subject lines.

> I have said this before, but repeat it too... I believe that not having
> optional constraints will lead clients to a style of programming that is
> likely to work only with specific devices, and may lead browser implementors
> to a style of programming where they ignore mandatory constraints with the
> argument that "they didn't know what they were asking for, we know better".

I'm not really sympathetic to this line of reasoning, especially since
it's so highly speculative.

On the other hand, if ignoring mandatory constraints turns out to be
the right thing to do, then it is the right thing to do.  I've heard
people say, with similar justification, that mandatory constraints are
a terrible idea.  Maybe they are right.

So if a future browser-maker determines that ignoring mandatory
constraints produces a global maximum for utility, then that is what
they should do.  Clearly, our predictions were bad.  And to the extent
that we, as specification writers, are made to look like fools in
retrospect, the end product is the same: a better user experience.

At that point, we change the spec to reflect reality and celebrate.
Until then, I'm not a fan of features that exist solely to assuage
some speculative fears.

Received on Tuesday, 26 March 2013 16:48:49 UTC