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

Re: AuthConfReq: Presentational Markup

From: Tab Atkins Jr. <jackalmage@gmail.com>
Date: Sun, 28 Mar 2010 10:20:41 -0700
Message-ID: <dd0fbad1003281020ya783e04s45e4079ae69880cb@mail.gmail.com>
To: Sam Ruby <rubys@intertwingly.net>
Cc: Maciej Stachowiak <mjs@apple.com>, HTML WG <public-html@w3.org>
On Sun, Mar 28, 2010 at 5:24 AM, Sam Ruby <rubys@intertwingly.net> wrote:
> The rationale provided listed a bunch of reasons why the presentational
> markup (e.g., the <font> tag) was disallowed, and I cited an example why not
> a single one of those reasons applied.  In response to Anne[1], I suggested
> that either the rationale was incomplete, or that the rationale needs to be
> challenged.
> I feel that is a fair thing to do.

Sure, and it's a fair thing to say "Yeah, that's an exception.  Shrug.
 Handling it properly would complicate the rules more than it's
worth.".  80/20 rule and all that.

> My central thesis is that banning is not the appropriate mechanism for
> markup that works interoperably and is widely and willfully used.  You are
> free to campaign against cheeseburgers, but are not free to outright ban
> their sale.
> Banning should have a rather high bar.  Any markup that is banned must have
> significant negative consequences and we need to be confident that any such
> use is not intentional.
> I (continue to) offer as an alternative the notion of identifying separately
> those notions that are felt to be Best Current Practices from those that are
> Author Conformance Criteria.

Pretty sure I see the central problem here.  You and I (and others)
are seeing the term 'non-conforming' as carrying different amounts of
weight.  It appears that you believe that decreeing a document as
non-conforming is a big step that should not be taken lightly.  I
believe that conformance is only relevant if you care about the
validator that's declaring you conformant or not.  If I felt that I
had good reasons to do something that the HTML5 spec deemed
'non-conforming', I'd double-check my reasoning, but then would do it
and wouldn't look back.  I'm confident that browsers will do the same
thing with my markup, so I can do things that aren't recommended if I
have to.

Frex, for a long time my old company's site used a layout table to
create a 2-column layout.  When I coded it I was perfectly aware of
the alternatives; in fact, before I did this, it *was* a float-based
2-col layout.  However, I decided that the existing CSS-based
solutions were too fragile, limiting, and complex for my needs, and
explicitly chose to use a layout table against semantics.  I wasn't
remorseful about this fact (though I did leave a funny comment in the
source about it), even though I was explicitly breaking a MUST
requirement in the spec.  It was just what I needed to do to ensure my
users had the best experience possible, while maintaining the visual
requirements of the site.  Since then, the browser ecosystem has
shifted sufficiently that I was able to switch to a proper CSS layout
using the table-* display values, and just feed a simplified
abspos-based layout to legacy browsers.

So, my site was non-conformant for about 2 years.  That's not a
problem - I *was* doing something bad.  I eventually changed it be
more conforming (at the moment it is valid HTML5, according to
validator.nu, and respects the author conformance requirements as far
as I know), when technology allowed me to do so without sacrificing my
user's experience.  The rules did what they were supposed to do, but
didn't get in my way when I decided to violate them.

I know you are aware of all of this, but you continue to sound as if
we're somehow striking things from existence when we forbid them, as
if the <font> tag suddenly disappears from the web or summons the
police when an author is detected typing it.  Being non-conformant
really isn't all that big of a deal.  A validator really is just such
a "goodness checker", as you put it in your most recent email, not a
sword hanging over one's head.

Received on Sunday, 28 March 2010 17:21:32 UTC

This archive was generated by hypermail 2.3.1 : Thursday, 29 October 2015 10:16:00 UTC