Re: Action-1634: the switch element and <title> and <desc>

Yes, the spec is wrong.  Or at least inconsistent.  It allows you to do
something that *looks* like it would be useful but is actually useless.

You can put a title and desc inside a switch, just like you can put them
inside any other container element.  However, you *can't* put a switch
inside a shape to switch between titles for that shape.  You can put a
switch inside a group, and titles inside that switch, but those titles
won't be associated with the parent group, they'll be associated with the
switch.  So a switch will never be used to select the title or desc for an
element.

And, since titles and descriptions aren't explicitly excluded from the
switch's switching rules, you can't use them to describe the switch as a
container element itself.  If you include a title as the first child of a
switch, that will be the child element that gets selected for the switch,
and all the other content within the switch construct will be ignored.  If
something else is included first, the title/desc will be ignored.  So a
switch will never have a meaningful title or desc itself; if it does, it
has no content to be named!

In other words, I should probably recommend that SVG 2 remove "descriptive
elements" from the content model for <switch>.

Amelia


On 18 May 2015 at 15:13, Richard Schwerdtfeger <schwer@us.ibm.com> wrote:

> Amelia,
>
> Perhaps I am reading this wrong but the <switch> element states that it
> can be a container for a range of different elements including "descriptive
> elements"
>
> When I follow the link for descriptive element i get:
> http://www.w3.org/TR/SVG2/struct.html#TermDescriptiveElement
>
> That text states the following:
> An element which provides supplementary descriptive information about its
> parent. Specifically, the following elements are descriptive elements: *‘*
> *desc* <http://www.w3.org/TR/SVG2/struct.html#DescElement>*’*, *‘*
> *metadata* <http://www.w3.org/TR/SVG2/struct.html#MetadataElement>*’* and
> *‘**title* <http://www.w3.org/TR/SVG2/struct.html#TitleElement>*’*.
>
> Is the spec. wrong?
>
> Rich
>
>
> Rich Schwerdtfeger
>
> [image: Inactive hide details for Amelia Bellamy-Royds ---05/16/2015
> 07:34:18 PM---Hi Richard, You're confusing two issues: language sw]Amelia
> Bellamy-Royds ---05/16/2015 07:34:18 PM---Hi Richard, You're confusing two
> issues: language switching for <title> and <desc>
>
> From: Amelia Bellamy-Royds <amelia.bellamy.royds@gmail.com>
> To: Richard Schwerdtfeger/Austin/IBM@IBMUS, public-svg-a11y@w3.org
> Date: 05/16/2015 07:34 PM
> Subject: Re: Action-1634: the switch element and <title> and <desc>
> ------------------------------
>
>
>
> Hi Richard,
>
> You're confusing two issues: language switching for <title> and <desc>
> versus content switching with <switch>.  The multi-language alt text
> options *won't* use <switch> because it wouldn't be backwards compatible.
> My response is therefore two parts...
>
> ____________________
>
> For the accessible text selection:
>
> The current text mostly looks good, except that the final sentence seems
> to be mangled.
>
> Here is a light edit:
>
>    Otherwise, unless the element is marked as presentational
>    (role="presentation" or role="none"):
>       - If performing a text alternative computation for an accessible
>       name and the current node provides a descendant <title> element chosen
>       based on the language rules for the SVG specification, return the <title>
>       element's text content as a flat string.
>       - If performing a text alternative computation for an accessible
>       description, and the current node provides a descendant <desc> chosen based
>       on the language rules for the SVG specification, return the concatenated
>       text and child content of that <desc> element.
>
>
> I've moved the "unless the element" bit to the top to make it clear that
> we're talking about the element being named, not the title/desc element.
> There should be no role on title/desc, and its obligatory role is none, but
> that doesn't prevent it from being used as the name for its parent.
>
> I've used "the concatenated text and child content of that <desc> element"
> because the SVG 1.1 specifications allow complex content within <desc>,
> including content from other namespaces.  This has been used in practice to
> include HTML markup within the description, although the specs don't
> specifically say what to do with it.  Feel free to replace with a better
> wording. It really should be the same wording as for aria-describedby;
> there is an editorial note in ACC-Name saying that they need to come up
> with explicit wording for aria-describedby, so we would want to pass our
> wording by them!
>
> __________________
>
> For <switch>:
>
> A <switch> changes which elements are rendered to the screen, so it should
> be addressed when building the accessibility tree.  A switch should never
> have any title/desc of its own.  The SVG 1 specs didn't specifically say
> that I can't have them, but they also don't exclude title/desc children
> from the switching mechanism, so including them never has the intended
> consequence!
>
> I would therefore lean towards making <switch> an obligatory role="none" /
> no role may be applied.  It's not quite presentational in the sense that
> using tables for layout is presentational, but it's only impact is on which
> other elements get displayed.  Does anyone think that is too strict?  It
> would mean ignoring any aria-label or other attributes.  Authors might add
> these if they are treating the <switch> as a group containing multiple
> presentations of the same content.
>
> As far as dealing with the switch behavior, that should go in section 5.1,
> which currently just references CORE-AAM (
> *http://www.w3.org/TR/core-aam-1.1/#mapping_general*
> <http://www.w3.org/TR/core-aam-1.1/#mapping_general>).  In that spec,
> there are rules for excluding or including elements in the accessibility
> tree.  I think we need a paragraph (possibly more informative than
> normative) about how those rules should apply to SVG: how they apply to
> non-rendered graphical elements, as well as how they apply to switch-ed
> content.
>
> I'll read over the CORE-AAM section and put together a draft text for the
> SVG doc.
> __________________
>
> There are two other SVG elements that I think need special treatment:
> <view> and <use>.  I'll send separate notes on them.
>
> Best,
> Amelia
>
>
> On 14 May 2015 at 09:51, Richard Schwerdtfeger <*schwer@us.ibm.com*
> <schwer@us.ibm.com>> wrote:
>
>    *https://www.w3.org/WAI/PF/Group/track/actions/1634*
>    <https://www.w3.org/WAI/PF/Group/track/actions/1634>
>
>    I am proposing a modification to the text of the accessible name and
>    description computation to better address the <switch> element. Please
>    review:
>
>    Here is the current text for step F of the SVG name and description
>    computation (*http://www.w3.org/TR/svg-aam-1.0/#mapping_additional_nd*
>    <http://www.w3.org/TR/svg-aam-1.0/#mapping_additional_nd>):
>
>    Otherwise, if performing a text alternative computation for an
>    accessible name and the current node provides a descendant <title> element
>    chosen based on the language rules for the SVG specification, return the
>    title text alternative as a flat string, unless the element is marked as
>    presentational (role="presentation" or role="none"). If performing a text
>    alternative computation for an accessible description, and the current node
>    provides a descendant <desc> chosen based on the language rules for the SVG
>    specification; return the description text alternative an accessible
>    description computation attribute as presentational (role="presentation" or
>    role="none").
>
>    Proposed Text:
>
>    Otherwise, if performing a text alternative computation for an
>    accessible name and the current node provides a descendant <switch> element
>    with one or more descendant <title>s return the title text alternative as a
>    flat text string determined by the switch, unless the element is marked
>    (role="presentation" or role="none"); otherwise if the current node does
>    not provide a descendant switch element, but it does provide a descendant
>    <text>, return the title text of the first descendant <title> element as a
>    flat text string, unless the element is marked (role="presentation" or
>    role="none"). If performing a text alternative computation for an
>    accessible description and the current node provides a descendant <switch>
>    element with one or more descendant <desc>s return the description text
>    alternative as a flat text string determined by the switch, unless the
>    element is marked (role="presentation" or role="none"); otherwise if the
>    current node does not provide a descendant switch element, but it does
>    provide a descendant <desc>, return the description text of the first
>    descendant <desc> element as a flat text string, unless the element is
>    marked (role="presentation" or role="none").
>
>
>
>    Rich Schwerdtfeger
>
>
>