[WEB4LIB] Re: 99.9% of web sites obsolete?

[Thomas Dowling]

At 09:13 AM 9/12/2002, Drew, Bill wrote:
>I am wondering about how standards are set my self.  How was decided that
>tables are "bad" and CSS is "good?"  What is wrong with browsers being
>forgiving of bad code in HTML?  I am asking these questions to advance the
>discussion.


The W3C establishes its standards through a process of reaching consensus 
among the consortium members, a group that includes the major makers of 
browsers and editors, among other groups.

A major W3C thrust for years is that document structure and document 
presentation should be described separately using languages optimized for 
each.  This goes back at least as far as the release of CSS1 in December 
1996 and was stated very explicitly in the December 1997 release of HTML 4.0.

There is no standard that describes tables as "bad."  They are obviously 
necessary for presentation of tabular data, and it is a pity that no 
browser has ever really supported to complete extent of table markup 
introduced in HTML 4.

What is "bad" - in the sense of "rightly deprecated because there are 
better ways to do it" - is using structural table markup in an effort to 
create a certain appearance or document presentation.  This is bad for 
accessibility reasons, especially if the table doesn't 
"linearize"  properly (that is, make sense when read in source-code 
order).  It is bad for usability of long pages because very few browsers 
will attempt to render any part of a table before having all of it to lay 
out.  The common use of pixel-based dimensions defeats resolution 
independence, and make unwarranted assumptions about the relationship 
between pixels and physical size.



What is wrong with browsers forgiving bad code in HTML?  IE5/Mac, IE6/Win, 
and Mozilla now all contain two separate HTML rendering engines, one for 
pages that are believed to use nearly standard markup, and one for pages 
believed to rely heavily on obsolete hacks.  Next time your browser is slow 
to open, think about how much time it's spending loading into memory a 
second engine that ought to be unnecessary.

For that matter, what exactly does it mean to "forgive" bad code?  Valid 
markup has defined meanings.  To the extent that markup slips away from 
valid syntax, it slips away from agreed-upon definitions.  At some point, 
the browser has to stop trying to guess what an author meant; anyone whose 
pages are one step beyond that point will complain that the browser should 
be more forgiving, but the line has to be drawn somewhere.


Thomas Dowling
OhioLINK - Ohio Library and Information Network
tdowling at ohiolink.edu