Dear Editors – Thank you for the feedback. I have a few comments.
1. This is not a sufficient technique for SC 2.4.6, which requires
that headings
and labels be descriptive, but does not require that they be present on the
page. As such, SC 2.4.6 addresses the text used for the heading or label.
Devarshi: Agreed. This should probably be listed as an advisory to Guideline
2.4. Any suggestions are welcome.
2. This could be an advisory technique, that is, a technique that is
not sufficient
for one of the success criteria, but which may make the contents more
accessible. However, it is still not clear which success criteria this
should be advisory to. Possibly to Success Criterion 2.4.10 (Section
Headings are used to organize the content) or to Guideline 2.4 (Provide ways
to help users navigate, find content, and determine where they are).
Devarshi: It can be an advisory to Guideline 2.4.
3. We have some concerns about using invisible text in CSS, especially with
ARIA landmarks on the horizon. The CSS for hiding them is not 100% stable.
Especially in right-to-left layouts, there are problems which left-to-right
layouts don't have. There are also ever changing decisions by screen readers
about what patterns they support. This technique would need to include the
technical details for how to hide the text using CSS.
Devarshi: Use of landmark roles and what this technique proposes are two
different things. Landmark roles are great, but their presence should not
undermine this technique.
I am not clear why this instability of text hidden using css is pointed out
here. Please note that the current w3 and wai pages use a h2 header called
<Site Navigation> hidden via css; refer to http://www.w3.org/ and
http://www.w3.org/WAI/. So far, it has worked, and more importantly without
users even noticing that this header was hidden.
Regarding the technical details needed, I will only be able to provide the
basic CSS required to hide the text.
Regarding issues with rtl layouts, I am not in favor of this technique
encompassing multiple design styles. This technique is for ltr layouts, and
if not including rtl will be a concern, a note stating this could be made in
the technique. I personally feel the most fundamental idea should be
conveyed, and then it is upon the designers and developers to supply a CSS
that hides text in an rtl layout, based on their understanding of this
technique.
Regarding differing screen reader patterns, I am not aware of this, not
since I started using them, though I cannot rule out changing user
patterns. Please clarify this point.
4. We also have concerns with requiring level 3 headings, rather than whatever
heading element works in the context of the page structure.
Devarshi: Agreed. This should be open to any header element.
5. The related techniques do not seem relevant to this technique, in that they
are not presenting alternative techniques for addressing this requirement.
Devarshi: G148 seems to be the most relevant. Any suggestions are welcome.
6. The test procedure should describe what needs to be true of the coding,
not how to test operationally. While we find it necessary to provide
operational tests for general techniques, they are less reliable than tests
that focus on the underlying technology. See the published CSS techniques
for examples of how to describe CSS tests.
Devarshi: Agreed.
Basically, except 3, the rest is fine with me.
Thanks,
Devarshi
Response from WG:
This could be an advisory technique for Guideline 2.4.
If you want to limit this technique to left-to-right layouts, please make that clear in the description.
Differing screen reader patterns would surface as accessibility support notes. You may wish to include a note similar to Note 1 in "C7: Using CSS to hide a portion of the link text". The technique of using CSS to hide text are very similar in C7 and in your proposed technique, and you may be able to borrow from other parts of that technique.
Fesponse from commenter:
1. This is not a sufficient technique for SC 2.4.6, which requires that headings and labels be descriptive, but does not require that they be present on the page. As such, SC 2.4.6 addresses the text used for the heading or label.
Devarshi: Agreed. This should probably be listed as an advisory to Guideline 2.4. Any suggestions are welcome.
2. This could be an advisory technique, that is, a technique that is not sufficient for one of the success criteria, but which may make the contents more accessible. However, it is still not clear which success criteria this should be advisory to. Possibly to Success Criterion 2.4.10 (Section Headings are used to organize the content) or to Guideline 2.4 (Provide ways to help users navigate, find content, and determine where they are).
Devarshi: It can be an advisory to Guideline 2.4.
3. We have some concerns about using invisible text in CSS, especially with ARIA landmarks on the horizon. The CSS for hiding them is not 100% stable. Especially in right-to-left layouts, there are problems which left-to-right layouts don't have. There are also ever changing decisions by screen readers about what patterns they support. This technique would need to include the technical details for how to hide the text using CSS.
Devarshi: Use of landmark roles and what this technique proposes are two different things. Landmark roles are great, but their presence should not undermine this technique.
I am not clear why this instability of text hidden using css is pointed out here. Please note that the current w3 and wai pages use a h2 header called <Site Navigation> hidden via css; refer to http://www.w3.org/ and http://www.w3.org/WAI/. So far, it has worked, and more importantly without users even noticing that this header was hidden.
Regarding the technical details needed, I will only be able to provide the basic CSS required to hide the text.
Regarding issues with rtl layouts, I am not in favor of this technique encompassing multiple design styles. This technique is for ltr layouts, and if not including rtl will be a concern, a note stating this could be made in the technique. I personally feel the most fundamental idea should be conveyed, and then it is upon the designers and developers to supply a CSS that hides text in an rtl layout, based on their understanding of this technique.
Regarding differing screen reader patterns, I am not aware of this, not since I started using them, though I cannot rule out changing user patterns. Please clarify this point.
4. We also have concerns with requiring level 3 headings, rather than whatever heading element works in the context of the page structure.
Devarshi: Agreed. This should be open to any header element.
5. The related techniques do not seem relevant to this technique, in that they are not presenting alternative techniques for addressing this requirement.
Devarshi: G148 seems to be the most relevant. Any suggestions are welcome.
6. The test procedure should describe what needs to be true of the coding, not how to test operationally. While we find it necessary to provide operational tests for general techniques, they are less reliable than tests that focus on the underlying technology. See the published CSS techniques for examples of how to describe CSS tests.
Devarshi: Agreed.
Basically, except 3, the rest is fine with me.