Can we make _urlForResource stop throwing runtime errors for resources that dont exist

[Jessica Forrester]

DataService get and list could both deal with a null/undefined/"" URL coming back by resolving its promise as an error.

But watch doesn't have a way to resolve something as error today, so if a watch failed on this issue there is nothing it can do to alert the controller.

[Ben Petersen]

Related to all the test noise?

DEBUG [phantomjs.launcher]: [OpenShift] _urlForResource called with unknown resource bogus/v1 [object Arguments]
DEBUG [phantomjs.launcher]:   :0 in apply
DEBUG [phantomjs.launcher]: [OpenShift] _urlForResource called with unknown resource pod/v1 [object Arguments]
  :0 in apply
DEBUG [phantomjs.launcher]: [OpenShift] _urlForResource called with unknown resource user/v1 [object Arguments]
  :0 in apply
  :0 in apply
[OpenShift] _urlForResource called with unknown resource clusterroles/unknown [object Arguments]
  :0 in apply
[OpenShift] _urlForResource called with unknown resource nodes/v1beta3 [object Arguments]
  :0 in apply
[OpenShift] _urlForResource called with unknown resource nodes/unknown [object Arguments]

This stuff is a little annoying as well.

[Jessica Forrester]

that comes from spec tests where we are validating that urlForResource does what we expect in the presence of unknown resources. So its expected currently. but yes we would still like to change how this behaves in general.

[Ben Petersen]

Ah, right its in DataService:

  DataService.prototype._urlForResource = function(resource, name, context, isWebsocket, params) {
    var apiInfo = APIService.apiInfo(resource);
    if (!apiInfo) {
      Logger.error("_urlForResource called with unknown resource", resource, arguments);
      return null;
    }

not the test. hmm.

[Sam Padgett]

Fixed by #304
/close