User-Agent: Mozilla/5.0 (Windows NT 6.0; rv:2.0.1) Gecko/20100101 Firefox/4.0.1 Build Identifier: Mozilla/5.0 (Windows NT 6.0; rv:2.0.1) Gecko/20100101 Firefox/4.0.1 If a table has a non empty summary attribute, the content of the attribute is used as the accessible name for the table. Reproducible: Always Steps to Reproduce: 1. go to http://www.html5accessibility.com/tests/table-thead-tfoot.html 2.use an inspect tool to view the acc name and acc description values for the 2 tables Actual Results: table 1: acc name is summary attribute content acc description is title attribute content table 2 acc name is summary attribute content Expected Results: table 1: acc name is caption element content acc description is summary attribute content table 2 acc name is caption element content acc description is summary attribute content may be that acc name should be concatenated text from title+caption for table 1 or only caption and title attribute content is discarded if both caption and summary present.

I do not believe that any screen reader currently does anything with the title attribute on table elements, not even those that parse the HTML rather than using the accessible tree. So I think if both caption and summary are present, we should discard the title. Questions: 1. If caption is NOT present, but summary is, is it OK then to use summary as the accessible name? 2. If caption is present, summary isn't, and title is, should we then use title as the source for accDescription?

This bug is confirmed.

I'm just worried that we might be rendering the caption twice, since the caption is a true element that gets its own accessible with its own accessibleName.

(In reply to comment #1) > I do not believe that any screen reader currently does anything with the > title attribute on table elements, not even those that parse the HTML rather > than using the accessible tree. So I think if both caption and summary are > present, we should discard the title. > > Questions: 1. If caption is NOT present, but summary is, is it OK then to > use summary as the accessible name? > 2. If caption is present, summary isn't, and title is, should we then use > title as the source for accDescription? answer to question 1: that's what JAWS does. answer to question 2: seems consistent with other acc name/description calculations no?

(In reply to comment #3) > I'm just worried that we might be rendering the caption twice, since the > caption is a true element that gets its own accessible with its own > accessibleName. IE 9 maps the the title (if present) to acc name and the caption to acc description. If no title is present, caption is sued for the acc name. doesn't do anything with summary.

Chrome 12 (windows) maps caption to acc name and summary to acc help doesn't do anything with title.

Steve, there's no spec or recommendations? We keep this behavior for years, we should be absolutely sure the new behaviour is a right thing.

(In reply to comment #7) > Steve, there's no spec or recommendations? > much of the implementation to the acc layer is not specced or agreed upon, this issue highlights the problem. > We keep this behavior for years, we should be absolutely sure the new > behaviour is a right thing. I wouldn't rush to chnage it.

1) caption as a name (bug 111284), compatibility with IE 2) summary as a name (bug 295715), caption is preferred over summary, Aaron's suggestion 3) summary as a name, caption as a description (bug 377783), Aaron's change, no motivation/discussion/objections I suggest to use # caption as a name; # use summary if caption is empty # use summary as description if not used as a name this logic is similar to title attribute processing. Sounds reasonable?

Additionally, if caption is empty and summary is becoming the name, and title is present, title becomes description.

(In reply to comment #10) > Additionally, if caption is empty and summary is becoming the name, and > title is present, title becomes description. sure it is (this is guaranteed by implementation).

Comment on attachment 541349 [details] [diff] [review] patch r=me thanks! Please test also that, if summary and title are present, but no caption, that title actually gets mapped to the accDescription. I'm missing that test part and would like to see it covered.

sure if you like

landed - http://hg.mozilla.org/mozilla-central/rev/63ce1a21bbd0