Re: ACTION-2107: Precedence of aria-details over aria-describedby

That is fine Matt. At least we are now on the same page. Your change will clarify what the group agreed upon. 

Thank you for your thoughtful review. 

Best,

Rich

> On Aug 16, 2016, at 11:33 AM, Matt King <a11ythinker@gmail.com> wrote:
> 
> Rich,
>  
> Thank you. I now understand that you interpret the text as placing a normative requirement on user agents to ignore aria-described if aria-details is specified. 
>  
> I am seriously not wanting to be a pain … but I do not see how you derive that requirement from the text. I do not see any normative statements about precedence. 
>  
> I will use your interpretation to complete action 2107. However, I do not think the spec is ready for CR if it has language we intend to be normative that does not specify whether it is directed at authors, user agents, or assistive technologies and does not include an RFC2119 must/should/may.
>  
> Now that I understand what the text means, I do not see the need for a requirement that makes aria-describedby and aria-details exclusive. 
>  
> I am not aware of other places in the spec where We use normative requirements to place this type of limitation on user experience. For example, we do not limit the number of elements that can be referenced by aria-laberlledby or aria-describedby. And, we support using those attributes to reference any mix of hidden and visible elements.
>  
> In the case where aria-details and aria-describedby are both specified, if the accessibility API can provide only one object relationship, I can see the need to specify that aria-details would be the one that has an object relationship revealed. In that case, the assistive technology would get the stringified aria-describedby content, but the available object relationship would refer to the aria-details. It would be important for the AT to know that the object relationship is for the details and not for the description. 
>  
> It could be the case that we are stretching the limits of APIs with aria-describedby, aria-errormessage, and aria-details. It is possible that all could be specified and refer to visible elements.
>  
> Matt
>  
> From: Rich Schwerdtfeger [mailto:richschwer@gmail.com] 
> Sent: Tuesday, August 16, 2016 4:11 AM
> To: Matt King <a11ythinker@gmail.com>
> Cc: ARIA Working Group <public-aria@w3.org>
> Subject: Re: ACTION-2107: Precedence of aria-details over aria-describedby
>  
> The text stated precedence, meaning that if both were provided one was chosen for the description. That text is in a normative part of the text. There was no statement that it MAY or SHOULD take precedence. It is black and white. 
>  
> Rich
>  
> 
> 
> Sent from my iPhone
> 
> On Aug 15, 2016, at 9:26 PM, Matt King <a11ythinker@gmail.com <mailto:a11ythinker@gmail.com>> wrote:
> 
>> Rich, if we do not put a normative statement for user agents in, then what would be the basis for following through with that in the AAM?
>>  
>> If that is the meaning of the text then we need something like:
>>  
>> If both aria-describedby and aria-details are provided on the same element, user agents MUST ignore the value of aria-describedby and expose only the value of aria-details.
>>  
>> Matt
>>  
>> From: Richard Schwerdtfeger [mailto:richschwer@gmail.com <mailto:richschwer@gmail.com>] 
>> Sent: Monday, August 15, 2016 7:42 AM
>> To: Matt King <a11ythinker@gmail.com <mailto:a11ythinker@gmail.com>>
>> Cc: ARIA Working Group <public-aria@w3.org <mailto:public-aria@w3.org>>
>> Subject: Re: ACTION-2107: Precedence of aria-details over aria-describedby
>>  
>> Matt, 
>>  
>> In the mappings we should suppress aria-describedby relationship mappings in favor of aria-details. That would also mean that aria-describedby content would not be converted to a string description. 
>>  
>> It is also possible that some mapping systems may reuse aria-describedby relationships and in those cases aria-details must win. From an author’s perspective that is very detailed low level information that will be system dependent. 
>>  
>> We foresaw that some platforms may reuse the same underlying linking (relationship) mechanism.
>>  
>> There must be only one description. The authoring practices should make should tell authors that only one description is allowed and that details takes precedence. 
>>  
>> If you want to write in the spec. that when aria-details and aria-describedby are provided on the same element the aria-describedby relationship will not be exposed to the AT that is fine for a clarification. 
>>  
>> Rich
>>> On Aug 14, 2016, at 10:22 PM, Matt King <a11ythinker@gmail.com <mailto:a11ythinker@gmail.com>> wrote:
>>>  
>>> Rich, 
>>>  
>>> I was out of the office during the recent discussion of aria-details. However, I did respond to the CFC regarding the normative changes with comments. Unfortunately, I was unexpectedly away again when the discussion of the CFC occurred and during which my comments led to action 2107.
>>>  
>>> I am not asking for a normative change in meaning because I don’t know the meaning of what is written. If you, or anyone else, can provide an explanation of how the precedence is created, enforced, or otherwise manifest, then I can complete the action of modifying the text to make it more understandable.
>>>  
>>> Is it possible that the group had discussed a precedence requirement but not actually made a decision of how it should be implemented? Could that be why there is not a normative statement placing a precedence requirement on either browsers or assistive technologies?
>>>  
>>> If there is not a normative implementation requirement, then removing the precedence statement is editorial. Birkir made an argument for not having a precedence requirement, and his rationale seems reasonable to me. But, again, I am making this judgment without having any understanding of the language stating there is a precedence. 
>>>  
>>> Matt
>>>  
>>> From: Rich Schwerdtfeger [mailto:richschwer@gmail.com <mailto:richschwer@gmail.com>] 
>>> Sent: Sunday, August 14, 2016 7:49 AM
>>> To: Matt King <a11ythinker@gmail.com <mailto:a11ythinker@gmail.com>>
>>> Cc: ARIA Working Group <public-aria@w3.org <mailto:public-aria@w3.org>>
>>> Subject: Re: ACTION-2107: Precedence of aria-details over aria-describedby
>>>  
>>> Ok. Now, I am having a serious problem. This was supposed to have been reviewed by the group and that includes you. It also went out for 7 day CFC. Where were you?  
>>>  
>>> Stating precedence is a normative statement. Do you think it is a nice to have? 
>>>  
>>> One of the problems I am now having with your comments is that if neither has precedence we have an overload of mechanisms a user must deal with to get help information. 
>>>  
>>> We need to have authors make a choice. The preference should be for everyone to be able to access the information and not the 1% of the users with an AT.
>>>  
>>> Remember this was put in to help digital publishers which is targeting all users.
>>>  
>>> Rich
>>>  
>>>  
>>> Rich
>>>  
>>> Rich
>>> 
>>> Sent from my iPhone
>>> 
>>> On Aug 13, 2016, at 8:23 PM, Matt King <a11ythinker@gmail.com <mailto:a11ythinker@gmail.com>> wrote:
>>> 
>>>> Rich,
>>>>  
>>>> The statement about precedence does not include any normative language. And, I still do not understand what it means. 
>>>>  
>>>> If both aria-describedby and aria-details are specified on the same element, is either the user agent or assistive technology supposed to do something special? If either or both are supposed to do something, what is it that they do?
>>>>  
>>>> Matt
>>>>  
>>>> From: Rich Schwerdtfeger [mailto:richschwer@gmail.com <mailto:richschwer@gmail.com>] 
>>>> Sent: Saturday, August 13, 2016 9:22 AM
>>>> To: Matt King <a11ythinker@gmail.com <mailto:a11ythinker@gmail.com>>
>>>> Cc: ARIA Working Group <public-aria@w3.org <mailto:public-aria@w3.org>>
>>>> Subject: Re: ACTION-2107: Precedence of aria-details over aria-describedby
>>>>  
>>>> That is true. However, we did not see the need for multiple descriptions. Also, if you change that it would be a normative change. 
>>>>  
>>>> Since both define a description, the reason for the precedence is that authors can hide aria-details content where they are not allowed to with aria-details - meaning it is accessible to everyone and not just AT users.  aria-details is intended to be shown. 
>>>>  
>>>> Rich
>>>>  
>>>> Rich Schwerdtfeger
>>>>  
>>>>  
>>>> 
>>>>  
>>>>> On Aug 12, 2016, at 4:56 PM, Matt King <a11ythinker@gmail.com <mailto:a11ythinker@gmail.com>> wrote:
>>>>>  
>>>>> WRT completing ACTION-2107, Make editorial changes to aria-details, I have one question.
>>>>>  
>>>>> What is the intended meaning of the following sentence from the aria-details specification?
>>>>> "When both aria-describedby and aria-details are provided on an element aria-details takes precedence."
>>>>>  
>>>>> Since aria-details is not part of the name and description calculation, it clearly does not refer to precedence in that calculation. Does it refer to a user agent behavior? If not, whose is responsible for creating the precedence, and how is that precedence manifest?
>>>>>  
>>>>> Thanks,
>>>>> matt