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

Re: css layout should be symmetrical

From: Vadim Plessky <lucy-ples@mtu-net.ru>
Date: Wed, 20 Feb 2002 11:14:24 +0000
To: "L. David Baron" <dbaron@fas.harvard.edu>
Cc: <www-style@w3.org>
Message-ID: <auto-000020137652@umail.ru>
On Tuesday 19 February 2002 18:41, L. David Baron wrote:
|   On Tue, 19 Feb 2002, Vadim Plessky wrote:
|   > Well, there is no warranty that Tables will be supported in future
|   > versions of CSS.
|   > So far, Tables is part of CSS3 Tables module - and this module is
|   > optional. You can develop CSS-compliant browser whcih doesn't support
|   > tables at all.
|   >
|   > Therefor, Tables in CSS *should not* be used for layout.
|
|   There's also no requirement that web browsers implement CSS.  So should
|   CSS never be used to suggest layout?  I would agree that pages using CSS

It depends on userAgent.
When someone develops some userAgent (Browser), he/she should determine:
* Target market
* system requirements
* typical applications used with this userAgent (Browser)
* typical platform for now or in nearest future

Don't get me wrong - it's up to developer to take those facts into 
consideration or not.
These are *Marketing* basics, every MBA graduate knows this.
But, if you look around and check different products available - those basics 
are rarely considered... :-(

I am pretty well aware that many browsers do not take into considerations 
typical platforms users have, and require too much (64MB of RAM, very fast 
HDD, a lot of disk space, etc.). And, none of major browser vendors can 
clearly explain what is their Target Auditory (Target Market) 
It pasrtially explains why despite a huge installed base those vendors failed 
to attract so-called "power" or "professional" users. 

If you check requirements for WAI (Web Accessibility Initiative), you will 
see that Tables are indeed not recommended to use for layouting, and CSS *is* 
recommended.
Of course it's again up to you to take WAI into consideration or not. I 
prefer to consider those recommendations. 

As about "wether to use CSS to layout or not":
I use Konqueror 99% of my time, and all layouting for HTML (and XHTML) is 
done via complimentary CSS stylesheets.
So, generally speaking, you can't disable CSS in Konqueror - I know this for 
sure, as some people were asking for it and answer was "no"

Is it possible to siable browser without CSS support?
Yes, of course. Look at NN and IE before 4.0 versions.
How valuable they are? it's a question.

But there is a market for a browser doing basic (I's say CSS3 Core - 
LineText, VisualBox model) CSS plus some subset of HTML (probably, XHTML 
Basic) and/or XML.
Most likely, you can get such browser into 512K of memory, or even less, and 
easily use in embeddd devices, appliances, etc.
Typical applications can be also mobile phones and PDAs.
You want tables being present, right?
Ok - than explain please why you may want to have CSS tables (and HTML  
Tables ) supported in your mobile or PDA. In both cases, *less is better* - 
so adding Tables is just overhead.

|   table display types should be designed to gracefully (to some degree)
|   degrade when those display types aren't supported, but that doesn't mean
|   those display types shouldn't be used if user-agent support is
|   sufficient.

It's a myth.
Tables *do not *gracefully degrade in old browsers (NN 1.x-4.x, IE 2.x-4.x), 
and all new browsers have more problems with broken table layouts in Netscape 
and MS IE than benefit form those Tables.
Check, for example, Mozilla - and measure how many bug reports were filed 
againt broken table layout?
And many of them still not fixed (I have Mozilla 0.9.8 and can confirm this)

|
|   (The box model module is just as optional as the tables module -- it
|   doesn't make sense on aural implementations, for example.  The
|   modularization (in my opinion) is part of a recognition of the reality
|   that it's difficult to implement all of CSS2 (and even more so for CSS3)
|   and that it should be possible to describe conformance criteria for
|   partial implementations.)

Fully agree!
|
|   > Tables in CSS should be used only with XML!
|
|   I disagree.  For many documents that contain user interface, the best
|   way to describe a layout using the existing features of CSS uses the
|   table display types.  Using floats to suggest complex table-like layouts
|   (as I've seen demonstrated) is a hack and is often somewhat fragile and
|   hard for authors to write.  The table display types certainly aren't
|   ideal -- I think a box model somewhat like XUL would be much better. [1]

I agree 100% that using float is a hack, ugly hack.
But tables is a hack as well.

I proposed some time ago to backport 'display: inline-block' from CSS3 to 
CSS2 - but so far I don't have _official_ answer from W3C WG about this, only 
few opinions, mostly supporting.
That is not perfect, but working solution.

Better solutions were also proposed on list (Tantek Celik, me, Jeffrey) - but 
so far we don't have official answer about them...

|
|   > again tables...
|   > We should get rid of them - tables are for represnting Tabular data,
|   > not for layouting!
|
|   The TABLE, TR, TH, TD, elements in HTML are for representing tabular
|   data.  The display types used to display something shouldn't need to
|   convey the meaning (semantics) -- the markup should already be doing

indeed. That's why I think that typical application for CSS2/3 Tables is XML.
You don't have default visual semantics for XML elements, you need to define 
that semantics.

|   that.  (FWIW, as a (perhaps typical?) native English speaker, I much
|   prefer "laying out" to "layouting".)

ok :-)
|
|   -David
|
|   [1] http://lists.w3.org/Archives/Public/www-style/2001Oct/0172.html

-- 

Vadim Plessky
http://kde2.newmail.ru  (English)
33 Window Decorations and 6 Widget Styles for KDE
http://kde2.newmail.ru/kde_themes.html
KDE mini-Themes
http://kde2.newmail.ru/themes/
Received on Wednesday, 20 February 2002 03:14:10 GMT

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