W3C home > Mailing lists > Public > public-html@w3.org > October 2010

Re: getting rid of callers

From: Boris Zbarsky <bzbarsky@MIT.EDU>
Date: Fri, 29 Oct 2010 17:29:44 -0400
Message-ID: <4CCB3CC8.3060107@mit.edu>
To: David Singer <singer@apple.com>
CC: robert@ocallahan.org, Ian Hickson <ian@hixie.ch>, "L. David Baron" <dbaron@dbaron.org>, Jonas Sicking <jonas@sicking.cc>, Cameron McCormack <cam@mcc.id.au>, public-html@w3.org
On 10/29/10 4:26 PM, David Singer wrote:
> I have no idea what the actual state here is, but plenty of companies
> have a policy of not removing features from shipping products unless
> it's really really necessary (e.g. a security fix), or after a *long*
> period of deprecation.

And that's fine.  And we want people to experiment with new web 
features.  That likely means putting the features in shipping products 
so they can be experimented with by web developers.

I think David's point is that just because that's happened doesn't mean 
the feature needs to immediately get standardized and required of all 
UAs, unless market forces are pushing that way anyway (e.g. the feature 
is being commonly used), in which case it's just that the experiment 
succeeded.

For new stuff things are somewhat mitigated in this area by using vendor 
prefixes, but even there never removing the prefixed feature can cause 
compat issues... yet we clearly don't want to standardize those.

For legacy failed experiments, which are of course unprefixed, I think 
we should apply the same sort of metrics. Likewise for non-legacy 
unprefixed things (e.g. new EcmaScript features which can't be done 
behind a vendor prefix).

-Boris
Received on Friday, 29 October 2010 21:30:24 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Wednesday, 9 May 2012 00:17:16 GMT