Re: CR exit criteria and features at risk for HTML5

On Sat, Aug 18, 2012 at 9:33 AM, Charles Pritchard <chuck@jumis.com> wrote:
> They must be broadly available and distributed. I personally have four
> implementations of HTML5 Canvas; I do not consider my personal quorum to
> meet the bar for interoperable implementations, despite their open source
> and provable interop.

To fix this, I think it should be required that each implementation be
in the context of a piece of software that's suitable for a person to
use as their primary means of browsing the Web.

This would exclude implementations that only contain one or a few
features of the spec but don't do so in the context of a piece of
software exposed to the full de facto compatibility requirements that
arise from having to be compatible with existing Web content to the
extent as to be suitable use as a person's primary means of browsing
the Web.

For example, I think the implementation of the HTML parsing algorithm
that I have written should be eligible for being used as an
implementation for assessing interoperability when considered as a
part of Firefox, but the Java version as a mere library or even
incorporated into an app that no one would use as their primary means
of browsing the Web should not be eligible for assessing
interoperability even though in the case of my example the mechanics
for deriving one of these implementations from the other ensure that
they run the exact same algorithm.

-- 
Henri Sivonen
hsivonen@iki.fi
http://hsivonen.iki.fi/

Received on Monday, 20 August 2012 08:21:16 UTC