Looking closer, why isn't the attribute mapped to aria-readonly="true"?

Good question. The mappings for aria-readonly=true appear a match for all platforms/APIs but AXAPI.

Currently the HTML-AAM has AXEnabled: NO for readonly, but "AXValue: the AXUIElementIsAttributeSettable method returns NO" for aria-readonly=true.

AXEnabled:NO seems wrong, and doesn't match what webkit is doing. But I can't confirm if the AXUIElementIsAttributeSettable method returns NO with AXValue. @cookiecrook or @fleizach, are we safe to map HTML readonly to aria-readonly="true" where AXAPI is concerned?

are we safe to map HTML readonly to aria-readonly="true" where AXAPI is concerned?

Yes. I believe so.

AXEnabled:NO seems wrong, and doesn't match what webkit is doing. But I can't confirm if the AXUIElementIsAttributeSettable method returns NO with AXValue.

Yeah, I agree that seems wrong. AXEnabled should map inversely to disabled/aria-disabled rather than readonly but perhaps some piece of the chain assumes that readonly is a superset of disabled? @fleizach may know.

bb106b9: Have updated all mappings for readonly to use CORE-AAM mappings for aria-readonly=true. If there are any issues, particularly with AXAPI (ping @fleizach?), can reopen.