On Jan 2, 2010, at 8:07 PM, Robert O'Callahan wrote:
> On Sun, Jan 3, 2010 at 3:42 PM, Maciej Stachowiak <mjs@apple.com>
> wrote:
> In the absence of any explicit attributes for buffer control, a
> likely good design would be to apply a heuristic. For example: if a
> page contains only one <video> element, then buffer aggressively. If
> it contains many, don't buffer any of them. Alternately, one could
> look at whether a particular video has larger explicit dimensions or
> appears in a more prominent place on the page. Since an unaware
> author is most likely not to add any special attributes, it would be
> nice to apply a heuristic like this when no special buffering-
> related attribute is present. Let's call this case (A).
>
> I'm unconvinced a heuristic is needed. Shouldn't we wait and see if
> authors actually fail to apply "autobuffer" when they should?
If we add a way for page to hint that no buffering is needed, we can
each act on our opinions as to whether a heuristic is needed, since
you can treat lack of attribute the same as the "no buffer" hint. But
if we don't add the third state, then either I can't add the heuristic
that I think is best for overall user experience, or authors have no
reliable way to opt out of buffering. It's true that, if you are
right, we end up with a no-op attribute, if we add nobuffer. But in
the meantime we could actually do the experiment of trying both ways.
Personally, I think it's highly likely that many authors will omit
autobuffer out of carelessness, and not as a signal of intent. It
would have no visible effect other than performance, and the
difference would be much less in the typical test environment where
the author has a very fast pipe to his or her deployment server. In
such circumstances, I don't see how we can possibly treat *lack* of an
attribute as a strong indicator of author intent. It's much more
likely to mean the author simply hasn't thought about the issue.
Regards,
Maciej