@DavidMacDonald I marked the example in your issue text as code because github was trying to render the image.

Surveyed https://www.w3.org/2002/09/wbs/35422/Oct_31_2017/

I support adding images of text.

Discussed on 10/26 call and almost came to a conclusion. https://www.w3.org/2017/10/26-ag-minutes.html

Two proposals discussed were:

• "For active user interface components with labels that include text, or images of text the name includes the text or the text in the image."
• "For user interface components with labels, the name includes the text of the label, including any text alternatives for images of text."

The main concern about the first proposal was that the second half of the SC allows either of the forms of text to be included in the name, but if both types are present then both need to be included.

The main concern about the latter proposal was that a text alternative for an image of text might have information beyond the equivalent of the text itself and the goal is to only require the text content, whether it is in an image or in text.

What about:
The name for a user interface component includes text that matches the content of any text and images of text used to provide a label for the user interface component.

@awkawk, I agree that the 2 proposals you quoted have issues, but I still don't understand what's wrong with the current pull request version (minus the "active"):

For user interface components with labels that include text or images of text, the name includes the text of the label.

That tries to say the same thing as your most recent offer, does it not? I think the complaint was that the second use of "text" could be taken confusingly since it refers to any text, regardless of whether or not it's an image. Perhaps that could just be clarified in a note or with a slight alteration like:

For user interface components with labels that include text or images of text, the name includes the text string as presented in the label.

I'd prefer keeping it clear that not all UI components have labels and not all labels have text or images of text rather than boiling that down to a single word ("any").

... or even "character string"

I think that the source of the problem is that if people are confused that text in a label doesn't necessarily include images of text why wouldn't they be confused about what they are required to include in the name?

How about:
"For user interface components with labels that include content that is visually represented as text, the name includes that content."

And then we provide a definition for "visually represented as text":
A visual representation of a sequence of characters where the sequence is expressing something in human language. The representation is not necessarily programmatically determinable.

I think that the source of the problem is that if people are confused that text in a label doesn't necessarily include images of text why wouldn't they be confused about what they are required to include in the name?

Well, yeah I agree but WCAG 2.0 shoots itself in the foot by defining "text" as being programmatically determinable, and then using "text" in the definition of "image of text" (straying from the dictionary causing problems sounds familiar...).

And then we provide a definition for "visually represented as text":

Do we really want to go down that road when we've already got definitions for text, image of text, non-text...?

How about just not using text in the second half and using the way it is defined as "sequence of characters":

For user interface components with labels that include text or images of text, the name contains the sequence of characters visually presented in the label.

I'm ok with that. Either way we will need to be clear in the understanding docs and have clarifying techniques.

Current proposal: For user interface components with labels that include text or images of text, the name contains the sequence of characters visually presented in the label.

Trying to simplify:

For user interface components with labels that include text or images of text, the name contains the text presented.

I.e. if you show text in whatever form, include it in the name. You could add 'visually' on the end, which would then exclude hidden labels, which would make sense for the voice-input scenario.

Version for CFC, as agreed by WG on 10/31/17 call: "For user interface components with labels that include text or images of text, the name contains the text presented."

Pull request #535 has been updated with the CFC wording.

Merged.