Resolve "Text has minimum contrast" feedback

[maryjom]

link to the survey results
This survey was discussed twice in meetings on:

• Initial discussion on 13 August
• Discussion of final survey results on 27 August

Kathy Eng

• I don't think 1.4.6 should be included since it requires minimum 7:1 (4.5:1 for large text) because the passing examples 2,3,4,5,6 would not pass.
• Passed Clarify what we mean by negative tests #7: Perhaps "Sample of Helvetica font" might be better than "quick brown fox..."? Either way, not certain this is "non-text content" and that aria-hidden should be true.
• Passed Clarify proof of compliance with absence of violation #8: Add "Note: Because this is non-text content, success criterion 1.4.11 Non-text Contrast requires font example to have a color contrast of 3:1." Contrast ratio is 3.6:1.
• I don't think this rule tests 1.4.6 adequately

Detlev Fischer

• Not sure about Passed Example 7:
  The example sentence won‘t show the aesthetics for LV users if the font if contrast is too low. So I would argue the sentence carries Information.

Wilco Fiers

• Not a blocker, but I wonder if we should recommend the not-a-widget part should be taken out?
• Remove the techniques from the mapping.

Mary Jo Mueller

• Not sure if all rules are continuing to have techniques listed in the mapping or not. If not, they should be removed.

[WilcoFiers]

Resolved everything except the question about the mapping to 1.4.6. I have a pull request to resolve that too: act-rules/act-rules.github.io#1460 Once that's merged, I think we can put this into final call. Don't think this needs another review round.

[WilcoFiers]

Ready for another survey

[maryjom]

Survey closed on 29 October.

[maryjom]

Discussing survey results at 12 Nov. meeting.

[maryjom]

Discussed survey results at 12 Nov. meeting.

Kathy Eng

• Rule assumptions: 1.4.6 is not mentioned
• Because 1.4.3 is AA and 1.4.6 is AAA, the implementations would be clearer if these to be 2 separate rules. Combining both in this rule may cause confusion on the passed implementations for readers (probably not for automated tools) since some fail 1.4.6.
• Similar situation exists for 1.2.3 (A), 1.2.5 (AA), and 1.2.8 (AAA). Combining all of these would be very confusing and this rule seems to set a precedence for combining them.
• Expectation only describes 1.4.3. Consider a separate rule for 1.4.6.

Mary Jo Mueller

• The assumptions are actually the same for both 1.4.3 AND 1.4.6, so I don't think that the specific criterion of 1.4.3 needs to be called out.
• Data looks correct, but so far we don't have any full implementations which blocks the rule from progressing through the process. There are some untested rules for SortSite and their tool passed Failed example 6. QualWeb also had a few inconsistent results, including Failed example 6. Is the shadow DOM something that automated tools cannot test? Or do these tools need updates to get the anticipated results?
• I'm assuming that this would be one atomic rule for 1.4.6 and another atomic rule would check the more stringent requirement in 1.4.6 - which renders mapping of 1.4.6 to this rule a bit moot. Why have 2 rules for 1.4.6 when you can have one?

Wilco Fiers

• We need to get at least 1 complete implementation. This rule will need to go on hold until we do.

Levon Spradlin

• I'm not sure what implementation data is lacking.

Trevor Bostic

• For inapplicable example 3, is there a reason to have the "aria-hidden='true'" attribute included? I think it confuses the point. I wonder if it would also be more clear if the description referenced the assumption as to why this counts as inapplicable. I am almost certain we had this discussion but I can't remember the reasoning. Why do we not consider text nodes that have ancestors with widget roles?