Re: Applying Postel's Law to HTML Data

On Thu, Oct 6, 2011 at 7:18 AM, Ivan Herman <ivan@w3.org> wrote:
> If we take those two statements above, I think what this means is that we should encourage authors to publish in one forma of their choice only.

That kind of choice for authors creates a long-term burden for all
consumers. For example, it's burdensome that content consumers need to
support application/xhtml+xml alongside text/html even though
application/xhtml+xml flopped. There's now enough
application/xhtml+xml legacy that removing support is scary.

I think we should want an outcome where there's one good way of
addressing a given use case. In the overlaid metadata context, this
means that for publishing metadata about a certain topic, there's one
framework and one vocabulary that everyone supports for metadata about
that topic.

If we don't know or disagree about which framework or which vocabulary
those should be, I think the path that allows wrong guesses to wither
away over time is the one where producers try many things
simultaneously instead of consumers supporting everything. In that
scenario, once we have winners, both producers and consumers can
support only those formats from then on.

Also, I think we should consider what kind of collateral damage and
legacy is left if a given format or the entire concept of overlaid
metadata flops. I think for development frameworks (e.g. browsers or
non-browser frameworks like the JDK) about the worst possible outcome
is that they implement support for many formats, then all of them flop
but not totally enough to allow support  to be removed and frameworks
have to maintain support for multiple format just for legacy
compatibility without the maintained code providing value in scenarios
involving new content.

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

Received on Thursday, 6 October 2011 12:02:01 UTC