Re: [webrtc-tests] Adding a basic no-media call that uses promises. (#1824)

I like tests that take the same code-paths that users would take. I see Harald's point in theory, but pragmatically, if the test ends up avoiding some race by being pedantic, then I see that as a disadvantage, not an advantage.

Also, to me, this pattern looks like a hold-over from the days when people sent offers from createOffer, something we now frown on since it's easy to check for failure on setLocalDescription with promises. I never did that in my tests, which is probably why the caching of the offer from createOffer (which excludes ice candidates that may have arrived in the time passed) seems so odd to me.

I don't want to belabor the point much further, as I don't know the scope of these tests well, whether they are meant to inform as examples or merely serve internal functions.

View on GitHub: https://github.com/w3c/web-platform-tests/pull/1824#issuecomment-103117959

Received on Monday, 18 May 2015 16:22:14 UTC