W3C home > Mailing lists > Public > w3c-wai-ig@w3.org > April to June 2002

Re: Frames and accessibility: opinions please

From: David Woolley <david@djwhome.demon.co.uk>
Date: Thu, 25 Apr 2002 23:01:05 +0100 (BST)
Message-Id: <200204252201.g3PM15105822@djwhome.demon.co.uk>
To: w3c-wai-ig@w3.org
> 
> Hi,
> 
> The new web site of the Dutch parlement http://www.tweede-kamer.nl has a 
> frame construction, without using the <FRAMESET> tag.
> As far as I can see in the source they use scripting to define the frame 
> structure.

No they don't.  They only use scripting to define the length of the contents,
to keep it within the bounds of the screen.  Even then, they are exceptionally
good for a commercially designed site in that they actually attempt to
provide <noscript> fallbacks.

> If I check this with Bobby it seems not to be aware, that this is a frame 
> constructions. Bobby give no priority 1 checkpoint error on missing titles.

That's because the titles aren't so important when you can see the whole
thing at once.  I've always assumed the requirement for titles is to get
round the problem where a frameset page displays as:

FRAME: top
FRAME: left
FRAME: bottom
FRAME: right 

and you have no idea which one to call up.

The use of frames has *never* been reccommended by W3C.  If their use had
been reccommended, their current state in HTML 4.0 and XHTML 1.0 would
be deprecated (their status in the next version of XHTML is deleted).
They only exist in the transitional versions that people should have
started phasing out 4 or more years ago.

The actual page actually works quite well in Lynx and IE5.5
with scripting disabled, although it has the typical tag soup
(e.g. <noscript><div></noscript>) and lack of alt attributes that one
comes to expect of commercial quality web designs.  The tag soup might
cause problems in browsers with different error recovery strategies.

> It is impossible to:
> - Navigate between frames;
> - A frame can not be loaded in a new window, which is often a solution to 
> access a page.

These too suggest recovery strategies from basically bad design.  With
specific reference to navigating between frames, a pure CSS design (this
looks like a hybrid tables CSS one) has the advantage that it can be
written to linearise cleanly, so can be read easily with simple tools that
don't attempt to honour the positioning at all.  Also, if you need to 
quickly access parts of the document, the good way of supporting this 
is to use Hn elements properly - I believe some AT tools will use these
to advantage.

Linearisation is essential if one is not restricted to designs which
are basically only suitable for large pixel addressable displays.

> - Most screen readers can not auto-page the content frame, so that 
> information outside the screen can never be accessed.

Sounds very much like a half baked implementation.  The simple approach 
with something like Lynx will linearise the document, and there should be
no problem - you can even put the navigation stuff to the end.  A full
implementation would know how to cope with CSS scrollable boxes.

I would rather have hoped that modern AT would access the document object
model, not the screen image (they more or less have to access it at some
level with GUI displays).

> - Opera 6 crashes on this construction if Use default is turned of in Page 
> layout settings.

Fix Opera if it crashes.  If it is simply confused by tag soup, blame the
tag soup (which is common to most commercial pages) not the frame like
layout.

> Last years AT Vendors did a lot of work to make frame constructions on the 

AT vendors have to work round the bad practice found in the wild; they
don't have the luxury of optimising designs for reccommended practice
(retire frames) if commercial designers insist on ignoring reccommendations.

> Internet more accessible and any purpose not using the standard <FRAMESET> 

FRAMESET is not a standard.  It is a transitional reccommendation to be
used until non-CSS browsers die out.  That transitional reccommendation is
either no longer the current reccommendation from W3C or will very soon
cease to be it.

Incidentally, you will find that all the first line portal sites have frame
like layouts (although they generally don't try and restict them to the
screen height) without using frames, even though many once used them.
Received on Thursday, 25 April 2002 18:08:43 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 19 July 2011 18:14:04 GMT