W3C home > Mailing lists > Public > public-html@w3.org > August 2007

Re: arguments for retention of LEGEND and FIELDSET

From: Maciej Stachowiak <mjs@apple.com>
Date: Sat, 4 Aug 2007 19:30:47 -0700
Cc: public-html@w3.org
Message-Id: <1F37E895-4A62-4EDC-8CE2-E2C812458BBD@apple.com>
To: "Gregory J. Rosmaita" <oedipus@hicom.net>

The fieldset and legend elements are present in the current draft of  
Web Forms 2 and are not deprecated. I have not heard of any plans to  
either remove or deprecate them, and I don't think the informative  
note you mentioned is in any way indicative of this. It describes how  
to find a form control's associated <fieldset> element by looking for  
it in the DOM ancestor chain, which is just the way it has always  

Has anyone said anything to the contrary?


On Aug 4, 2007, at 6:58 PM, Gregory J. Rosmaita wrote:

> in the Web Forms 2.0 section 7.11 it is not only written:
> quote
> 7.11. Labels
> Form controls all have a labels DOM attribute that lists all the
> label elements that refer to the control (either through the for
> attribute or via containership),in document order.
> Similarly, HTMLLabelElements have a control DOM attribute that
> points to the associated element node, if any.
> A label must be listed in the labels list of the control to which
> its control attribute points, and no other.
> unquote
> then, in a cavalier aside, it is noted:
> quote
> Assistive technologies may use the labels attribute to determine
> what label to read out when a control is focused. An assistive
> technology could also wish to determine if the element is in a
> fieldset group. To do so, it should walk up the element's
> parentNode chain to find the fieldset ancestors.
> unquote
> this is insufficient and an enormous step backwards...
> why?  because an assistive technology, such as a screen reader,
> functions quite like a blind person, not knowing what it is
> which it has come into contact with, UNLESS that item has an
> explicit LABEL, belongs to an explicit FIELDSET, whose title
> is encased in LEGEND.
> reasoning:
> 1. an assistive technology cannot be relied upon to correctly
>   infer a FIELDSET; FORM controls MUST be explicitly
>   contained in a FIELDSET;
> 2. a FIELDSET contains a related grouping of form controls,
>   each one of which needs to be individually labeled.  the
>   LEGEND allows a non-visual visitor's assistive technology
>   to contextualize the FORM controls bound to the LEGEND,
>   by virtue of their inclusion in the FIELDSET,  there is
>   no way for an assistive technology to associate the
>   nearest header with a form control grouping, unless that
>   header also serves as the LEGEND for the FIELDSET of FORM
>   controls that allow one to "Reply to this Comment".(that,
>   and headers are important for structure and navigation,
>   and it is fitting to use a LEGEND to encase a header.
> 3. what is needed is an explicit, not implicit, FIELDSET,
>   and where there is a FIELDSET, there must be a LEGEND.
> 4. when used in a TABLE-ized FORM, an explicit LEGEND allows
>   assistive technologies to associate the LEGEND with the
>   FIELDSET, although anyone who uses TABLE to control
>   presentation is misusing an element that deserves
>   deprecation in HTML5, as a TABLE has meaning only insofar
>   as one can perceive the visual relationships between data
>   and its labels and categorizations.  therefore, it is
>   merely a presentational model, better handled - and more
>   appropriately relegated - to CSS.
> 5. an element should NOT be deprecated due to incompenent
>   unimaginative authoring practices (relying on a browser's
>   default styling for LEGEND and FIELDSET) and/or incomplete
>   implementation on the user agent's part; (this was given to
>   me as a reason why Web Forms 2.0 would be dropping LEGEND
>   and FIELDSET because of default browser styling; this
>   constitutes a failure of imagination and implementation on
>   an author's part, and CAN be handled QUITE easily using
>   the CSS box model)
> gregory.
> ----------------------------------------------------------------
> CONSERVATIVE, n.  A statesman who is enamored of existing evils,
> as distinguished from the Liberal, who wishes to replace them
> with others.         -- Ambrose Bierce, _The Devil's Dictionary_
> ----------------------------------------------------------------
>             Gregory J. Rosmaita, oedipus@hicom.net
>  Camera Obscura: http://www.hicom.net/~oedipus/index.html
> ----------------------------------------------------------------
Received on Sunday, 5 August 2007 02:30:56 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 16:25:10 UTC