W3C home > Mailing lists > Public > public-aria@w3.org > May 2018

Sanity checking one of the name calculation test cases

From: Joanmarie Diggs <jdiggs@igalia.com>
Date: Wed, 9 May 2018 15:33:41 -0400
To: ARIA Working Group <public-aria@w3.org>
Message-ID: <bfe215a8-575e-9bf0-9da8-1f87a7217713@igalia.com>
Hey all.

Name test case 663 [1] has expectations that seem a) beyond our spec and
b) potentially wrong. The test case is:

if given
  <style type="text/css">
    label:before { content: "foo"; }
    label:after { content: "baz"; }
    <label for="test">
      <input id="test" type="file" name="test" title="bar">
then the accessible name of the element with id of "test" is "foo bar baz"

User agents are including the "foo" and "baz". But they are not
including "bar". Instead, they are including the text displayed on the
button, such as "Browse...".

My thoughts:

The behavior I'm seeing is consistent with what the HTML-AAM states for
other buttons (like submit and reset). [2] Mind you, the HTML-AAM
doesn't explicitly address the file input button scenario, so I've filed
an issue against that spec so that they can fix that. [3]

Back when I initially saw this behavior/failure with test case 663, I
created a test that was the same but used an input type of "image"
instead of file. That is test case "663a" [4]. It passes. So we have
coverage for the general scenario.

I believe testing what happens with the file input should be done as
part of HTML-AAM testing; not Accname testing. That, plus the fact that
we have coverage for the general scenario in 663a, has led me to
conclude that the correct thing to do is remove this test from our
CR-exit-criteria suite. Any objections?


[3] https://github.com/w3c/html-aam/issues/134
Received on Wednesday, 9 May 2018 19:34:18 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 9 May 2018 19:34:19 UTC