1141695 - write functional tests for 2way media connection direct calls telemetry

Status: RESOLVED WONTFIX

(Hello (Loop) :: Client, defect)

Description

[Dan Mosedale (:dmosedale, :dmose)]

+++ This bug was initially created as a clone of Bug #1141690 +++

Bug 1123660 included functional/integration tests that verify the following behavior:

* link generator creates a room, 2-way media start ("media-start") is NOT noted
* link clicker enters the same room, media start IS noted
* link clicker leaves the room, call length IS noted (even though window is still open)

And bug 1141690 describes these functional/integration tests:

We'd like to do more calls to cover that the following cases all note the call length in the appropriate sequence (e.g. that local click hangup doesn't wait for window close to note the call)

* local click hangup (disconnectSession)
* local close window (disconnectionSession)
* remote close window (RemotePeerDisconnected with peerHangup==true)
* remote click hangup (RemotePeerDisconnected with peerHangup==true).
* remote click network failure (onConnectionDestroyed, RemotePeerDisconnected with peerHangup=false)
* local network error (onSessionDisconnected, actions.ConnectionFailure), disconnection: note END)

There's probably a spin-off bug one could file here for similar coverage of the above cases where the link generator leaves a room, and then the clicker enters it first.

The trick is going to be limiting scope (the 6 cases listed here would be at least 6 separate calls).

If we really want to do a bunch more calls, we want to avoid the case where functional tests take so long to run, that people don't get the advantage of using them in the development process directly, so I suspect these calls might want to be in a separate suite that gets run by the automated testing machines, but not by default in a desktop development environment.

Comment 1

[Dan Mosedale (:dmosedale, :dmose)]

To be clear, the above list is from the rooms bug.  This bug is for morphing that into the analogous direct call cases, and implementing the most critical ones here.

Comment 2

[Dan Mosedale (:dmosedale, :dmose)]

A good test to start with is likely to be automating something along the lines of this comment:

https://bugzilla.mozilla.org/show_bug.cgi?id=1145819#c11

Note that a fair bit of that automation is likely to be written in bug 1141694, which this bug depends on.

Comment 3

[Mark Banner (:standard8)]

Direct calls have gone away now -> wontfix.