W3C home > Mailing lists > Public > www-style@w3.org > January 2002

Re: conformance (was layout solutions blah blah blah)

From: Etan Wexler <ewexler@stickdog.com>
Date: Fri, 25 Jan 2002 08:20:03 -0500 (EST)
Message-Id: <20020125131928.20B416353@server1.safepages.com>
To: "Tantek Celik" <tantek@cs.stanford.edu>, Web style list <www-style@w3.org>
>> I am sorry to bear this message, but neither Internet Explorer
>> 5 Macintosh Edition nor Internet Explorer 6
>> conformantly implements CSS1.
> 
> Last time we checked, both fully passed the CSS1 test suite, 
the only
> objective measure of CSS1 conformance that exists.

Conformance criteria are clear in CSS1 section 7.  There is 
no mention in that section that the test suite is the measure 
of conformance.  There is no correction or 
supplementary Recommendation that defines the test suite as 
the measure of conformance.

If a product passes the test suite, the vendor should state as 
much, and with pride.  But I decline to confuse conformance 
with passing an incomplete test suite.

> So, I'd say that leaves the burden of proof in the naysayers' court.

I agree.  However, I do not want to clutter www-style with 
the details of implementation bugs.  That enumeration belongs on 
a developers list, on Usenet, or in the Quality 
Assurance departments of the vendors.

> Implementing CSS2 and CSS3 does not invalidate
> CSS1 conformance by design.

I refer to CSS1 section 7:

A User Agent that uses CSS1 to display documents
conforms to the CSS1 specification if it:
 * attempts to fetch all referenced style sheets
   and parse them according to this specification
 * sorts the declarations according to the
   cascading order
 * implements the CSS1 functionality within the
   constraints of the presentation medium (see
   explanation below)."

The criterion is not "parse them according to this specification 
or according to later versions of CSS".  At the lexical level, CSS1 
and CSS2 are largely but not entirely compatible.

CSS1 accepts only characters in the ISO-8859-1 repertoire and 
can refer only to codepoints in the Basic Multilingual Plane.  
CSS2 accepts any codepoint in the Unicode range, directly or 
by reference.

The selector "fo\6F bar" is equivalent to "foo bar" according to 
CSS1 and is equivalent to "foobar" according to CSS2.

And so on.

>> Conformance is not the soul and lifeblood of Web software.
> 
> Strongly disagreed. It is exactly that.

Let me qualify my statement.  For some purposes (for 
example, business transactions), conformance is crucial.  For 
the majority of Web users, conformance is not even a factor.

>> Most people, myself included, value speed, stability,
>> and convenience as much as if not more than they
>> value conformance to specifications.
> 
> This was perhaps true in the old WWW (Wild West
> Web) of the mid 1990s, but is no longer true
> today in the 2000s.

Are we living on the same planet?  Most people are not aware of 
the *existence* of formal specifications for the Web, let alone of 
the details of those specifications or of the conformance 
of implementations.  I offer an example conversation:

Everyman: "What are you doing there, Etan?"
Etan: "Cascading Style Sheets."
Everyman: [blank stare]
Etan: "You know, CSS?"
Everyman: [blank stare]
Etan: "A computer language for styling structured documents."
Everyman: [blank stare]
Etan: "It makes Web pages look pretty." [grumble]
Everyman: "Oh, Web pages!  I know Web pages!"

Seriously, though, are you telling me that people would use a 
CSS1-conformant browser if it offered no printing, saving, or 
text extraction facilities; if it had no navigational history and 
no bookmarks; if each glyph required two seconds to render; and 
if the browser crashed every few minutes?

> The bar has been raised.  People expect valid
> pages to appear correct.  New browsers that
> fail this criteria have been severely
> criticized for it.

> http://www.alistapart.com/stories/omniweb/

Do not mistake Daniel "waferbaby" Bogan or Jeffrey Zeldman 
for the Great Unwashed Masses.  The only thing that most 
people expect on the Web is that pages render in 
unfamiliar browsers as they render in the browser of 
personal choice.  And most people will converge on a single browser.

What percentage of Web pages validate according to 
formal specifications?  Well, then, what level of demand should 
we anticipate for software that does the Right Thing, when 
the majority of Web content is tag soup?  (Confer "The Rise 
of 'Worse is Better' '' <http://www.ai.mit.edu/docs/articles/goo
d-news/subsection3.2.1.html>.)

-- 
Etan Wexler
Received on Friday, 25 January 2002 10:06:58 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 27 April 2009 13:54:12 GMT