memory leak on trials with reward

[ransona]

There memory leak that seems to be associated with the new version of bonsai or new workflow and causes a crash after about 20-30 stimuli. If we run the following to generate 30 trials with the same grating the memory usage gradually creeps up until the computer crashes. As you see I tried putting a rig.clear() in the loop after each stimulus is requested but this makes no difference. As you can see this happens over a few mins and depends upon actually showing stimuli - i.e. if you pause stim requests the memory usage stops going up.

Minimal example with python attached.

readamransonandneurogearsltd.zip

[ransona]

update: on the old workflow / old bonsai version if you start a gratings trial with rig.go(100,100,100,1) instead of the passive rig.start() then you also get the memory leak. I guess that by making passive trials rewardable in the latest version you have caused it to have this behaviour with passive gratings.