webrtc / utilities Goto Github PK

Since travis-multirunner only supports linux (debian based) we could skip if Mac and Windows are detected and use already installed browsers.

This is better than not being able to run anything at all.

This is already possible today but it would be nice if we could do it automagically instead of having to modify the test framework code every time.

AppRTC
Samples
TestRTC
adapter.js

not sure if I should file it here or in testrtc... I noticed some issues while trying to use it:

1. it uses the video elements width/height here
   which is not set for apprtc. This should be videoWidth/videoHeight, right?

2. it listens for the play event which means it can't be attached to video which already play:
   https://github.com/webrtc/testrtc/blob/master/src/js/camresolutionstest.js#L307
   I think the right thing is to listen for play only when the readyState is not 4/HAVE_ENOUGH_DATA and start checking right away when it is in state 4. @samdutton, you know more about this than I do :-)

cc @andresusanopinto

I've made quite a number of changes in adapter's selenium-lib recently which should be ported over to here. Notably https://github.com/webrtc/adapter/pull/258

This library is not really needed anymore. The custom test runner only works for Firefox and Chrome on Linux. The eslint configuration can be replaced with a standard one instead.

https://github.com/samccone/issue-and-pr-templates

Description
When debugging issues that only occur on travis it would be really helpful if we could grab a screenshot of the error and make it view-able somewhere.

• Add screenshot handling in src/js/selenium-lib.js using Webdriver api's (does not look like this is part of the Javascript api, need to investigate this further)
• Store the file on the travis bot and then upload it S3, documentation here: https://docs.travis-ci.com/user/uploading-artifacts/
• Include the link to the screenshot in the travis output