ACT Rules ready for approval (November 2023)

[WilcoFiers]

The ACT Task Force has completed work on the following three rules, and request approval from the working group to publish these on the W3C website. If you have comments, please respond to the thread.

Also note that ACT Rules are atomic tests. They are not intended to test complete success. If you are not familiar with ACT Rules, please first read about ACT Rules and the ACT Rules format 1.0.

Please vote 👍 if you approve of the change, or leave a comment if you have concerns.

[WilcoFiers]

Meta element has no refresh delay (no exception)

This is an Atomic rule to test 2.2.4: Interruptions (Level AAA), 3.2.5: Change on Request (Level AAA)

This rule checks that the meta element is not used for delayed redirecting or refreshing.

[GreggVan]

good to add a plain language explanation of what the element does. "If this is non-zero the page will automatically refresh after that time has expired - which can interrupt a user unexpectedly" (or whatever is a better, more complete, or more accurate description of the meta element)

[WilcoFiers]

Opened an issue to go with this: act-rules/act-rules.github.io#2151

[WilcoFiers]

HTML images contain no text

This is an Atomic rule to test 1.4.5: Images of Text (Level AA), 1.4.9: Images of Text (No Exception) (Level AAA)

This rule checks that images of text are not used

[WilcoFiers]

I tried rewriting the assumption to see if this could be clarified, see: https://github.com/act-rules/act-rules.github.io/pull/2149/files#diff-92587515eaeca8f5e156032c3003c940d71e9033a1bbf3eebe11b72f8b545d96R48

And yes your understanding is correct. This rule should not be used on pages where the text rendering in images can be adjusted. This is not too different from how most ACT rules shouldn't be used on pages with a conforming alternative version.

[maryjom]

When I read the text, "This rule assumes that there is no mechanism to change the rendering of text in image resources on the page, and that with CSS any requires presentation is achievable." Should "requires" be "required"?

[GreggVan]

We should mention that (alternate version comment) in the test itself (and in all tests where it would apply). Note this is different than an alternate accessible page or section -- but just to places where the actual item itself can be made accessible. That would seem to be a more limited set. -- and we should give an example of how this could be met.

[PoornimaSubramanian]

One suggestion is that we might want to bring out the different types of images e.g. SVG, JPEG, GIF and mention which one is in impact for this rule for more clarity. Because HTML image sounds more generic and all types of images go into the HTML markup.

[WilcoFiers]

@GreggVan This is a general principle of ACT Rules. It is explained in the about ACT rules page. This is also exactly how WCAG criteria work. A page can conform even if a criterion fails if that page has a conforming alternative version.

@PoornimaSubramanian Great suggestion! I've opened an issue here: act-rules/act-rules.github.io#2152 Certainly its worth mentioning that formats like TIFF aren't supported in all browsers, and so can result in some variability.

[WilcoFiers]

Heading is descriptive

This is an Atomic rule to test 2.4.6: Headings and Labels (Level AA)

This rule checks that headings describe the topic or purpose of the content.

[GreggVan]

if the visual and tree order are different -- and that is not allowed or a failure of something else -- that should be mentioned in this rule -- rather than saying that it should be evaluated using the tree not the visual order. They should be the same and isn't it a bad idea to suggest that they would be different? it should be "testable using either the visual or the tree since they would need to be the same or they would fail xxxxxx"

[JAWS-test]

Why "Perceivable content" is defined as "if the node is an element, it does not have a semantic role of none or presentation."?

<h1>heading</h1><p role=none>part one</p><p>part two</p>

Does the heading then have to match the second paragraph and not the first? If so, why?

[JAWS-test]

<h1>What is a descriptive heading?</h1>

<p>Hi, friends. It's Sunday and the sun is shining. It's 15 degrees. Last week it was always cloudy and cold. 
So why am I sitting at the computer instead of going out and enjoying the nice weather?

<p>Let me explain. I have an important topic. But I didn't have time for it last week. 
Last week I had to deal with <a href="https://github.com/w3c/aria-practices/issues/2888#issuecomment-1858849956">
menus</a>.

<p>But now I'm getting to the topic. I'm already in the third paragraph and only now do I realize 
how the heading fits this text. I'm talking about the ACT rule "Heading is descriptive"

<p>According to this rule, the heading must match the first content, e.g. paragraph, in the text. 
Do you see the problem? In the first paragraph, I only talk about the weather. 
So would a heading "The weather" be better? Hardly. Because the weather only serves as an introduction 
to my topic, but I'm actually writing about headings here. If my heading was "The weather", 
the wrong people would read my text. Because the text is about descriptive headings.

<p>Ok, now you could say that my text is not a purely technical text, but more like an essay, 
because I don't get straight to the point, but ramble on about the weather and the last week. 
An essay is almost literature - and literature and headings is a very difficult subject.

<p>Take a poem. There are supposed to be poems where the meaning of the heading is only revealed 
in the last line - and only for the few gifted people who are able to understand the poem. 
This makes me wonder why there is no exception for literature in SC 2.4.6, but that is a 
different issue, as the ACT Rule is not responsible for this shortcoming.  

<p>Does every site that publishes texts by Shakespeare have to adjust all the headings 
so that finally the heading matches the content or even the first paragraph. Hopefully not. 
But now let's move on to a purely technical text. Second attempt, 
without weather and pointless introductions. 

<h1>What is a descriptive heading?</h1>

<h2>The Middle Ages</h2>

<p>The Roman Empire reached its greatest territorial extent during the 2nd century AD; the following 
two centuries witnessed the slow decline of Roman control over its outlying territories. 
Runaway inflation, external pressure on the frontiers, and outbreaks of plague combined 
to create the Crisis of the Third Century, with emperors coming to the throne only 
to be rapidly replaced by new usurpers. Military expenses steadily increased, 
mainly in response to the war with the Sasanian Empire. The army doubled in size, 
and cavalry and smaller units replaced the legion as the main tactical unit. 
The need for revenue led to increased taxes and a decline in numbers of the curial, 
or landowning, class. More bureaucrats were needed in the central administration to deal 
with the needs of the army. ...

<p>...

<h2>Two headings that don't fit?</h2>

<p>What do headings have to do with the Middle Ages and what do the Middle Ages 
have to do with the time before the Middle Ages?

<p>Quite simply: imagine a book that explains the Middle Ages. But in order to understand 
the Middle Ages, I have to explain prehistory. So the first paragraph is not about the Middle Ages, 
but the second or third is. 

<p>And imagine I write a text about meaningful headings and use the example of the Middle Ages. 
Exactly what you are seeing here. Are my two headings wrong again? I don't think so.  

<h2>Conclusion</h2>

<p>I'm coming to the end now because, as I said, the sun is shining and I'd rather be outside:

<p>I don't think the rule that a heading is only descriptive if it describes the first content 
after the heading makes sense. Please think about it again. 

<p>And by the way, does at least the last heading fulfill the rule? I have no idea. 

<h2>Sources</h2>

<p>The text about the Roman Empire comes from the Wikipedia entry on the Middle Ages

[JAWS-test]

And who actually says that the first "Perceivable content" after the heading is content at all, which must somehow be meaningfully related to the heading. It can also be

• the name of the author
• the date
• a button to print the text
• a link to create a bookmark for the text
  etc.

[WilcoFiers]

Thanks for the feedback everyone. I see understand the hesitation about 1. Using the accessibility tree here instead of visual order, and 2. the idea of only checking the first content after the heading. We'll discuss this in ACT TF to see what we can do about this.

[WilcoFiers]

Deprecate the xml:lang matches lang rule

Next up, we are proposing to deprecate the HTML page lang and xml:lang attributes have matching values rule. This rule is no longer relevant as all modern browsers now prefer lang over xml:lang in all situations.

[WilcoFiers]

Restructure spacing rules

The spacing rules have been restructured. When testing these rules it is now required to ensure text is actually impacted by the style attribute. Empty elements no longer fail this rule, nor do nested elements where the spacing override is unset.

• Important line height in style attributes is wide enough
• Important word spacing in style attributes is wide enough
• Important letter spacing in style attributes is wide enough

[JAWS-test]

Understanding:

Content does not fail this SC if the method chosen by the user - for instance, the use of an extension or bookmarklet - fails to correctly set the line height and spacing text properties on the content (provided that the content is not actively and purposely preventing the properties from being added).

I'm not sure if this passage is enough to assume that the use of important in the style attribute is forbidden. It could also be considered a bug of the tools if they are not able to override the author's styles. The bookmarklet that is used for testing in Germany, for example, overwrites the important information in the style attribute. I would like to get feedback from more people here

[WilcoFiers]

Agreed. This criterion is difficult to test against. It largely depends on the capability of whatever tool(s) you want to support. That's true for lots of assistive technologies though. Screen readers have huge differences too. This is more basic though. These rules are written with "Stylus"-like technology in mind. If that's what you're aiming to support, these rules work. If all you're after is support for custom stylesheets (I think Safari is the only browser that has those), those can override !important attributes, in that case these rules aren't the right way to test the SC.

We tried to put some information on this in the accessibility support section of the rule, but ideally this SC's understanding document would explain this better.

[gundulaniemann]

I might miss the overall picture of the rules concenring spacing, yet I miss paragraph spacing in the set of changed rules.
I do not understand the change description:
"When testing these rules it is now required to ensure text is actually impacted by the style attribute. Empty elements no longer fail this rule, nor do nested elements where the spacing override is unset."
ensure the text is actually impacted means, that the width really changes.
Yet what is tested now is that the space is wide enough if it cannot be overridden.
It might be my Englisch, yet I do not understand the phrase 'where the spacing override is unset'. Can you please explain?

Do I understand correctly that the rule does not ensure that the SC is still passed when the font size is changed? In some passing examples, the spacing is set in px, so when the font size is enlarged, the spacing is no longer larger enough.

Why is inapplicable example 7 inapplicable?

[WilcoFiers]

I might miss the overall picture of the rules concenring spacing, yet I miss paragraph spacing in the set of changed rules.

We don't currently have a rule for paragraph spacing. That shouldn't prevent us from maintaining existing rules though I think.

It might be my Englisch, yet I do not understand the phrase 'where the spacing override is unset'. Can you please explain?

Have a look at this example: https://www.w3.org/WAI/standards-guidelines/act/rules/24afc2/proposed/#inapplicable-example-6 Here the letter-spacing is set on the p element, but the span inside it overrides that p's letter-spacing.

Do I understand correctly that the rule does not ensure that the SC is still passed when the font size is changed? In some passing examples, the spacing is set in px, so when the font size is enlarged, the spacing is no longer larger enough.

I'm not sure I understand this. Because px is not a value relative to the text (unlike em for example) you can have a px value that is sufficient for small text, but insufficient for large text. Does that answer your question?

Why is inapplicable example 7 inapplicable?

Do you mean of the line-height rule, because it has a style element with !important in it? This unlike when done with an attribute, a style element, even where the rule has !important in it, can be overwritten.

[WilcoFiers]

Other minor changes:

• Iframe with interactive elements is not excluded from tab-order
  Added inapplicable example 5 and 6

• Menuitem has non-empty accessible name
  Updated accessibility support section

• Element with lang attribute has valid language tag
  Update example descriptions of failed example 4 and 5

[WilcoFiers]

RESOLUTION: The following are approved by AGWG:

• Meta element has no refresh delay (no exception)
• HTML images contain no text
• Deprecate the xml:lang matches lang rule
• Restructure spacing rules
• Other minor changes

The following is NOT approved and will be revisited by the ACT Task Force:

• Heading is descriptive