Fork me on GitHub
#planck
<
2019-07-25
>
leonoel16:07:25

I'm having a travis failure for a test suite that works locally https://travis-ci.org/leonoel/missionary Does the dump at the end look familiar to someone ?

mfikes17:07:09

@leonoel I think it might be possible to get Planck to crash in odd ways if it shuts down while there are threads processing things. (It shouldn't crash, of course, but I think I've seen this before.) So, my first hunch is that there might be a timer in your script that is firing but the script ends, and things derail. Does that sound like it might be a possibility?

mfikes17:07:42

Oh, cool, I see your code is open source.

leonoel17:07:45

the test suite is full of timers

leonoel17:07:48

using cljs.test/async

mfikes17:07:56

@leonoel FWIW, it is passing for me locally with lumo

$ clojure -Acljs-test -x lumo

Testing missionary.core-test

Ran 17 tests containing 0 assertions.
0 failures, 0 errors.
But when I run it locally with planck, it terminates prior to the tests completing.

mfikes17:07:04

I smell a Planck bug.

mfikes17:07:48

(To run with lumo, use 3.7.0 of the test runner.)

leonoel17:07:27

good to know

mfikes17:07:06

Sometimes it runs properly, but sometimes it just terminates

Mikes-iMac:missionary mfikes$ clojure -Acljs-test -x planck

Testing missionary.core-test

Ran 17 tests containing 0 assertions.
0 failures, 0 errors.
Mikes-iMac:missionary mfikes$ clojure -Acljs-test -x planck
Mikes-iMac:missionary mfikes$

mfikes17:07:32

Planck will specifically keep the process running if it knows there are outstanding timers.

mfikes17:07:47

(Just like Node.)

leonoel17:07:21

OK I will try to investigate more. Thanks for your time

mfikes17:07:54

@leonoel I can see that the code in Planck that blocks until timers is complete isn't actually blocking. Seeing if I can sort out what that bug might be.

👌 4