Re: NOFRAMES is the right way.

> I think we should advocate the use of NOFRAMES for all FRAMESETs. Although
> by and large the UA can solve the problem for simple sets, it is not
> really the best solution - the best solution is to solve it by design, and
> the responsibility is on the author to do that.

I agree an author can achieve a better job by adding a NOFRAMES
section by hand, but if a few UAs can do it in an accessible way, I
still reluctant to ask millions of authors to redo their design
manually, when the responsibility really lies into the hand of the
language designer (the people that designed FRAME without careful
thinking).

There is something purely visual about FRAME: they are primarily meant
to save the back-and-forth navigation between an index and a content
page by keeping the index *visible* at all time.

In a serial media, such as voice or line braille, you cannot really
provide this functionality, and a new design is required. With the
tools and languages we have today, the UA can do an accessible but not
a perfect job at this new design.

To consider a very simple case, if an author writes

FRAMESET
  FRAME vertical left 20% name=index
  FRAME leftover name=content

then the UA can easily provide a degraded interface made of links to
the individual FRAME:
  [index]
  [content]

whereas the NOFRAME author would probably have designed a start page
with the index directly inline, thus saving a intermediary link page
to the serial reader.

But consider a new language construct an author can use to designate
the "entry-point" page in its frameset (index in my example), an
"variable-content-page", "fixed-page", etc, then suddenly, the UA can
do a much better job, approaching closely what the author would do in
NOFRAME, without duplication.

I guess my point is that it's for WAI PF to think about, and not for
GL to start asking authors about.

But I'm OK with P2 for all FRAMESET if that's what people want.

Received on Tuesday, 2 February 1999 04:59:11 UTC