Re: Feedback on SC1-2-2 from autowcag meeting 25 September 2014

Op 2 okt. 2014, om 16:10 heeft kamyar <kamyar.rasta@tingtun.no> het volgende geschreven:

> 
> 
> 
> -------- Forwarded Message --------
> Subject: Feedback on SC1-2-2 from autowcag meeting 25 September 2014
> Date: Tue, 30 Sep 2014 16:16:11 +0200
> From: kamyar <kamyar.rasta@tingtun.no>
> To: Hanno.Lans@Kinggemeenten.nl, wilco Fiers <w.fiers@accessibility.nl>,
> mikael Snaprud <mikael.snaprud@tingtun.no>
> 
> Hi Hanno,
> Here is the discussion on tests that you have made,
> SC1-2-2-video-element-without-track and SC1-2-2-video-element during the
> autowcag meeting on 2014-09-25. This is a compilation of comments on the
> discussion page of each test and descriptions of comments and new
> suggestions.
> 
> Best Regards,
> Kamyar
> 
> ------
> Attendees:
> Annika, Birkir, Kamyar, Wilco
> 
> Review 1.2.2 Video without tracks:
> https://www.w3.org/community/auto-wcag/wiki/SC1-2-2-video-element-without-track
> 
> Most comments are in the discussion page on the wiki already:
> https://www.w3.org/community/auto-wcag/wiki/index.php?title=Talk:SC1-2-2-video-element-without-track&oldid=581
> 
> - Step 1 should be manual. There is no aria role to specify content as a
> media alternative to text.
> - This page doesn't use the same logic flow as the test on for example
> this page: https://www.w3.org/community/auto-wcag/wiki/SC4-1-1-tag-nesting
> - A few of the steps do contain outcome tables but don't explicitly
> state when to return those outcomes (although most of the time it is
> clear from the text). For example "For each spoken audio fragment the
> frame contains text" (step 4) could be
> - Testmode is missing on each step and selector.
> - "IF NOT for each spoken audio fragment the frame contains text: RETURN
> {{{id}}}"
> --Wfiers
> 
> - Suggestion: Move the manual steps that check open captions to a
> separate test. This test could be applied to other videos as well (not
> just video elements).
> - The test mode of the steps should be either automatic or manual.
> SemiAuto should not be used (in step 1 and 2). In this case the step
> should be spilt into two steps.
> --Anietzio
> 
> Birkir: We may add a separate technique for flash to automatic detection
> of flash videos. The flash objects needs to be asked if they contain
> video. Wilco: there is no any technique to detect videos. However, video
> players can identify them.

I personally think we shouldnt invest in flash techniques. html5 video is the standard. You will also see that websites include a script with an automatic fallback or upgrade to/from html5 with all videoplayers: if the browser can handle html5, the video will be served as html5 (JW player, Youtube etc). If we test in a HTML5 compatible environment, this script will serve us HTML5 VIDEO and not an OBJECT. 
If we want to test the fallback methods, we have to test in an HTML4 as well as a HTML5 environment.


> 
> SUGGESTION: Extent this test to object element. Look at video and object
> elements to capture all of the possible variation of it.

If we want we can add a test for objects containing a video file as a parameter. When they have a video file as a parameter we might assume that the object is used for video
NB: if there is an object without a video file, it could be a video as well.
NB2: the file might be a videocontainer with a caption stream.

For me I think that as soon as we detect a OBJECT we should return a can'ttell, manual testing needed for this successcriterion.
So, I can include an extra text for objects.

> 
> -----------------------------
> 
> Review 1.2.2 Video element:
> https://www.w3.org/community/auto-wcag/wiki/SC1-2-2-video-element
> Comments on discussion page:
> https://www.w3.org/community/auto-wcag/wiki/index.php?title=Talk:SC1-2-2-video-element&oldid=579
> 
> - This page doesn't use the same logic flow as the test on for example
> this page:
> https://www.w3.org/community/auto-wcag/wiki/SC4-1-1-tag-nesting
I have added a few IF ELSE and GO TO statements
> - A few of the steps do contain outcome tables but don't explicitly
> state when to return those outcomes (although most of the time it is
> clear from the text). For example "File content contains time codes and
> text" (step 4) could be
> "IF File content does't contains time codes and text: RETURN
> SC122-videoelement-fail#"
> - Testmode is missing on each step and selector.
> - The page states "success criterion doesn't take into account language
> changes. So it is valid to have spoken text in language 1, subtitles in
> language 2 and page content in language 3." But I would argue that in
> that case the captions are not an equivalent to the audio track.
> --Wfiers

I don't understand when we have to return a pass or a fail because of the language. 
Let's assume we have a German website, that contains a video Obama visiting Merkel.

Obama and Merkel are talking in English.
- Is it required for WCAG2 to have the spoken text translated in subtitles in the German language?
- Is it required for WCAG2 to have captions in the German language?
- Is it required for WCAG2 to have the audio description in the German language?

> - Suggestion: Add reference to Technique H95
> - Selector: There is not track-attribute in HTML5. It should say: "track
> element with kind attribute".
Agreed
> - Step 2: Not clear what happens if the attribute is found.
> 
> Something is missing from description. Maybe it should say “go to the
> next step”. Not clear do we treat subtitles different from captions...

That's also not clear for me. I would argue that when there is no additional non-spoken information needed, only a subtitle file is needed, and only if additional clues are needed, a separate caption file is needed. Or is a caption file always required?....

> Suggestion: Step 1 and Step 2 could be merged to the selector because
> there is no difference in how captions and subtitles are treated.

We should discuss if we treat them the same
> 
> - Step 8: What is the intention of this step? Why is there no need for
> further testing for 'talking head' videos?
> Suggestion: Remove this step.

There is an exclusion in wcag2 especially for talking head videos. I included this as this test could be done automatically with image recognition and we can skip the other steps. It is a precondition test.
When it is not a talking head video we have to deal with checks for background, 

> - Step 10: What is a "background fragment"? Step 11: What is a "speaker
> fragment"? Please clarify.

Hmm, yeah, we need better definitions. Any suggestions to specify these situations?

> - Suggest to split this test into two parts. (1) automated part checking
> for the presence and correct format of the file. (2) Manual check that
> the captions describe the video.

How could we split a test in two parts in the format?

> - The test mode of the steps should be either automatic or manual.
> SemiAuto should not be used (in step 6 and 8). In this case the step
> should be spilt into two steps.
ok

> --Anietzio
> 
> SUGGESTION: remove the graded rating scheme from these tests.
> SUGGESTION: We do not include synchronized media in to selector. We just
> select video in object element, we can automate that, otherwise we have
> to ask user to verify every page with synchronized media.


> 
> 
> 
> 

Hanno Lans www.datascape.nl
Middenweg 73, 2024 XA Haarlem
06-26076205 | hanno@datascape.nl | @hannolans 

Received on Thursday, 16 October 2014 14:11:54 UTC