Closed scotthavens closed 4 years ago
There are some odd problems with the list copy for the thread variables. Might have to use the bigger deepcopy
guns...
I think that there are some thread variables that are persisting between test runs, that was the problem I was having especially going from station runs to HRRR runs.
I think that there are some thread variables that are persisting between test runs, that was the problem I was having especially going from station runs to HRRR runs.
Curious on why that only happens on the Python 3.6 runs. Last night, I was trying to see if there were some changes in Python and could not find any. I will have another look. Locally they consistently pass with 3.8
It's not just the 3.6, I've seen it pass there then fail on OSX 3.8. So there is something that is persisting between the simulations in the queue
I did some more reworking of the thread variable handling https://github.com/USDA-ARS-NWRC/smrf/tree/18_threading
to prevent convoluting the code with deepcopy
I think we should close one of the branches. My vote would be for your fork, so we can both update.
Sounds good, this one is puzzling!
Closing in favor of #178
I just updated the comparison for the time variable and it now passes. I am a little suspicious of that success though. Can you run it on your local for a double check?
Trying to tackle the larger issue #114. Added a specific Lakes test that uses gridded data and threading. There were a few things that needed to be fixed and I have moved all the thread variables to their modules for better control. Also added tests for outputting specific variables from both the threaded and non threaded case.