- From: cathychan <web-platform-tests-notifications@w3.org>
- Date: Fri, 04 Apr 2014 15:37:05 GMT
- To: public-web-platform-tests-notifications@w3.org
The permission model being used in these tests is different from the one depicted in the spec. The spec suggests implicit consent through user interaction, whereas the test instructions presume an explicit allow/deny prompt ("Click the button, allow camera access, and start to capture an image.", "Make sure that there is no granted camera access.", "Click the button, deny camera access.")
Furthermore, the tests do not verify that the UA correctly produces a file when the user goes ahead with a capture, or correctly produces no files when the user denies the capture (implicitly by cancelling the capture).
I would re-work the capture tests as follows, using capture_image.html and capture_image_deny.html as examples.
capture_image.html
- instructions: Click the button, and capture an image.
- add script to verify that one (and only one) image file becomes available.
capture_image_deny.html
- instructions: Click the button, but cancel capturing an image.
- add script to verify that no files become available.
I would skip the check on whether an image capture screen is shown, or put that in a separate test, as that is actually a SHOULD requirement.
View on GitHub: https://github.com/w3c/web-platform-tests/pull/306#issuecomment-39578557
Received on Friday, 4 April 2014 15:37:12 UTC