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

Re: Decentralised extensibility idea (ISSUE-41)

From: Aryeh Gregor <Simetrical+w3c@gmail.com>
Date: Sun, 17 Jan 2010 07:10:53 -0500
Message-ID: <7c2a12e21001170410y4bc49671lf5067c9bcb6af51a@mail.gmail.com>
To: Leif Halvard Silli <xn--mlform-iua@xn--mlform-iua.no>
Cc: Toby Inkster <tai@g5n.co.uk>, Philip J├Ągenstedt <philipj@opera.com>, "public-html@w3.org" <public-html@w3.org>
On Sun, Jan 17, 2010 at 12:56 AM, Leif Halvard Silli
<xn--mlform-iua@xn--mlform-iua.no> wrote:
> You know, if you think so, then you should work for removing @data-*
> from the spec. And don't forget @class either. Because both of them
> allows to sneak across the borderline without your attention.

It's always *possible*.  You could always shoehorn nonstandard data
somewhere.  You could hide it in the least significant bits of in-page
images if you wanted.  But at least the obvious extension points
should have the "applicable specification" requirement added, so that
at those extension points, widely-used validators will only validate
things they know are are open standards.

In practice, it seems like authors of language extensions don't want
to hide in class/title/etc. (Microformats notable excepted).  It's
cumbersome and inflexible.  So you can either use a nasty hack (data-*
makes it only slightly less nasty) and validate, or have much more
freedom but not validate unless you standardize.  That's better than

> All that can sneak through a validator with @profile can also sneak
> through it without @profile. @profile can not be used for blessing
> something that isn't already part of HTML5. @profile is not something
> which lets you invent your own HTML elements and attributes.

Why is your proposal preferable to inventing your own elements and
attributes?  As far as I can tell, the only advantage is it encourages
people to try sneaking past validators with extensions that haven't
been verified as open standards.  IMO that's a *bad* thing.

> @profile is for creating open, well-conceived standards. Microdata as
> well, btw. I have no clue how Microdata validation will work, but I
> thought - in addition to certain centrally kept vocabs - it was meant
> for individually developed vocabularies. Even the W3 and the WHATwg
> would not be powerful enough to keep track of all the "applicable"
> vocabularies that could be created for Microdata.

"The item type must be a type defined in an applicable specification."

This means that yes, microdata vocabularies must be whitelisted by
validator authors to validate.  They don't have to be approved by the
W3C or WHATWG, though.  If a community develops that independently and
responsibly manages microdata vocabularies and has a list that's
clearly specified, I'd expect validators to whitelist all the
vocabularies on that community's list.

Of course, vocabularies will still *work* even if they don't validate,
as always -- you can still *use* them just fine.  They just won't get
the imprimatur of a trusted validator until they are, in fact, open
standards.  (Assuming that trusted validators are pro-standards, which
has historically been the case.)
Received on Sunday, 17 January 2010 12:11:28 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 16:25:42 UTC