SHould be able to do it based on postedMessages again - I just noted that you currently send an error response if the pendingResolution doesn't exist - I'm changing that to return a success respons...
The important thing is that it has disconnected then reconnected, but state got cleaned up in between. Hence the challenge is simulating the disconnection - I think just calling the cleanup functio...
window.open was actually fine (at least in my tests), it returns a window object synchronously which can be compared with other window proxy objects. It was iframes that were the problem, as you ha...
@robmoffat is your best bet for some questions.
> At the moment when the tests are running we have app resolver UIs opening in the root container (if our agent gets an intent request with no app...
Yes, cleanup was not being performed prior to this (when reconnect was fixed, it generated an additional copy of each broadcast or intent event for each reconnection). I guess that's a case of chec...
Its so we can check fully realized windows first before waiting on anything to be realized. That matters in `getInstanceForWindow` where you are comparing window objects, we want to compare to all...
nope eslint still flags it. Apparently the _ syntax is dated an current thinking is that we should remove unused variables, so the linting tools flag regardless of an _ prefix.
Elsewhere, I've c...
The heartbeatHandler is in charge of goodbye messages and timeouts, so yes should probably call this - or what it does should be moved into setAppState. The tests don't exercise heartbeat disconnec...
When I found the issue I addressed with this, the state for the app was already gone - I couldn't quite figure out why. Possibly some issue in `setInstanceDetails` (which replaces the connections a...
Yes on this one, it's a defensive `else` clause that will give a meaningful error if someone breaks the implementation. For now it's not possible to reach but a better option than a non-null assert...
Created an issue for 2.3 to plug that gap: https://github.com/finos/FDC3/issues/1490
I'll see if I can drop in a test to just check it throws an error, we then check the error message when a sta...
There's no standard behaviour to test here - there is no error defined for this case. Any test we put in will be enforcing arbitrary behaviour so I didn't see any value in either tracking or testin...
## Enhancement Request
As discussed at meeting #1478, the `DesktopAgentProxy` implementation used by getAgent to communicate with bowser-based Desktop Agents currently uses 10 second timeout for...
## Describe your change
Testing CI check when run on a PR from a repository fork (which are failing at the last step, posting a comment with test results)
@novavi This is also ready to go and implemented in [fdc3-for-web-getAgent-refactor](https://github.com/finos/FDC3/tree/fdc3-for-web-impl-getAgent-refactor). THe handling is implemented in: https:/...
@novavi this is implemented in [fdc3-for-web-getAgent-refactor](https://github.com/finos/FDC3/tree/fdc3-for-web-impl-getAgent-refactor) via functions in this file: https://github.com/finos/FDC3/blo...