This is a continuation of the conversation we had on whether or not the autocomplete attribute value needs to be used with a specific type attribute value (as per the HTML spec)? The question is raised in the context of sc 1.3.5 Identify input purpose.
This issue is closely related to #1541 and I don't think we can submit autocomplete attribute has valid value for a review until we resolve it.
I've created two [test pages for the autocomplete attribute]. One for autocomplete with a correct control group and one for autocomplete with an incorrect control group. The findings are:
- Firefox is providing suggestions for both
bday-month and username regardless of the type;
- Chrome is providing suggestions for
username regardless of the type and does not provide any suggestions for bday-month
- Safari behaves just like Chrome
In summary, it looks like the browsers do not care about the type and do not provide suggestions/ autocompletion for all autocomplete attribute values.
The situation is slightly complicated given that sc 1.3.5 does not require autocompletion but for the property to be programmatically identifiable. If the rule was purely about autocompletion then I'm sure we should stick with the spec but because using autocomplete for 1.3.5 has somewhat become a secondary purpose of the autocomplete, I'm not so sure the rule should be so strict on the correct type attribute part, especially given that browsers do not seem to care.
It would be good to have a discussion about this during the next call.
This is a continuation of the conversation we had on whether or not the
autocompleteattribute value needs to be used with a specifictypeattribute value (as per the HTML spec)? The question is raised in the context of sc 1.3.5 Identify input purpose.This issue is closely related to #1541 and I don't think we can submit
autocompleteattribute has valid value for a review until we resolve it.I've created two [test pages for the
autocompleteattribute]. One forautocompletewith a correct control group and one forautocompletewith an incorrect control group. The findings are:bday-monthandusernameregardless of thetype;usernameregardless of thetypeand does not provide any suggestions forbday-monthIn summary, it looks like the browsers do not care about the
typeand do not provide suggestions/ autocompletion for allautocompleteattribute values.The situation is slightly complicated given that sc 1.3.5 does not require autocompletion but for the property to be programmatically identifiable. If the rule was purely about autocompletion then I'm sure we should stick with the spec but because using
autocompletefor 1.3.5 has somewhat become a secondary purpose of theautocomplete, I'm not so sure the rule should be so strict on the correcttypeattribute part, especially given that browsers do not seem to care.It would be good to have a discussion about this during the next call.