namespace_browser: Possibly fixing the flakyness.
I believe this is the cause:
Tests assume, rightly so, no state mutation happens after
the end event. hasChildrenPromise listens to data event
and mutates a state (item.isGlobbable.set(true);) and cancels
the context, but if a new data comes in before RPC is cancelled
chances are end event has already fired and now we are mutating
state again. Simple switch to .once() shoud fix this.
Also:
-Realizing Promise.all does not do what I expected in
some cases and replacing it with Bluebird.settle (although
based on the current state of code .all and .settle should
behave the same, technically .settle is the right API to use)
-Renaming streamError to globError
-Making tests print more when they fail
Change-Id: I3e7afd7a001b22394a8dc1eff997f1ec21237f44
diff --git a/package.json b/package.json
index 2aba86a..1709f2b 100644
--- a/package.json
+++ b/package.json
@@ -32,6 +32,7 @@
"routes": "^1.2.0",
"lodash": "~3.0.0",
"vis": "~3.9.1",
- "extend": "~2.0.0"
+ "extend": "~2.0.0",
+ "bluebird": "~2.3.2"
}
}
