How to debug no message matching on assert_receive - testing

Question

I have this test
defmodule InfoSys.Backends.WolframTest do
use ExUnit.Case, async: true
alias InfoSys.WolFram
test "make request, report results, then terminates" do
ref = make_ref()
{:ok, pid} = WolFram.start_link("1 + 1", ref, self(), 1)
assert_receive {:results, ^ref, [%InfoSys.Result{text: "2"}]}
end
end
and I am receiving
No message matching {:results, ^ref, [%InfoSys.Result{text: "2"}]}
after 100ms.
How can I know which message is ref receiving or how can I debug this error? I am following phoenix book programming example
Whole ExUnit stacktrace:
1) test make request, report results, then terminates (InfoSys.Backends.WolframTest)
test/backends/wolfram_test.exs:6
No message matching {:results, ^ref, [%InfoSys.Result{text: "2"}]} after 100ms.
The following variables were pinned:
ref = #Reference<0.214527998.4009754625.219099>
The process mailbox is empty.
code: assert_receive {:results, ^ref, [%InfoSys.Result{text: "2"}]}
stacktrace:
test/backends/wolfram_test.exs:11: (test)

Answer 1

I've ran on these kind of problems before. There are two probable causes:
Your process is taking longer than 100 ms to give you the result.
Your process has a bug.
I prefer to rule out the first possibility and the thing I do is making the test wait more than the default time which is 100 ms. assert_receive can receive a second argument which is the time in milliseconds. You can try something like this:
assert_receive {:results, ^ref, [%InfoSys.Result{text: "2"}]}, 5000
where 5000 means 5000 ms or 5 seconds.
After increasing the time, there are two possible outcomes when running a test like this:
Your test mailbox is not empty: your process is slow when processing your requests. Nothing wrong with the logic. Maybe the thing your process is doing is an expensive operation or you need to improve your code's efficiency.
Your test mailbox is empty:
There is definitely a problem with the logic of your process and you
should check the code.
You have underestimated the complexity of the operation your process should do and you should find a more efficient way of doing it or change your solution.
I've seen the code you are taking about and it's not too complex (next time you should post it as part of the question), so if you don't receive a response in your test after 5 seconds, then there is definitely something wrong with your process' code.
I hope this helps.