Use cloudpickle for serializing NegLogParameterPriors

[dweindl]

Cloudpickle is able to handle more complex objects than pickle.

See #1465
Closes #1465

[codecov-commenter]

⚠️ Please install the to ensure uploads and comments are reliably processed by Codecov.

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 82.83%. Comparing base (ab61c4a) to head (48e4743).

❗ Your organization needs to install the Codecov GitHub app to enable full functionality.

Additional details and impacted files

@@             Coverage Diff             @@
##           develop    #1467      +/-   ##
===========================================
- Coverage    82.87%   82.83%   -0.05%     
===========================================
  Files          163      163              
  Lines        13760    13765       +5     
===========================================
- Hits         11404    11402       -2     
- Misses        2356     2363       +7     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[PaulJonasJost]

Does it make sense to perhaps test with specifically a model where it previously failed (e.g. Schwen_PONE2014 from #1465)?

[dweindl]

Does it make sense to perhaps test with specifically a model where it previously failed (e.g. Schwen_PONE2014 from #1465)?

Generally, yes. I already have that lying around. But it will be another slow test due to model import, that's why I left it out so far.

[PaulJonasJost]

Does it make sense to perhaps test with specifically a model where it previously failed (e.g. Schwen_PONE2014 from #1465)?

Generally, yes. I already have that lying around. But it will be another slow test due to model import, that's why I left it out so far.

As we always have the problem to balance between "elaborate" test vs time. Would it make sense to have a set of tests that only run in the (I think weekly) master test?

[dweindl]

As we always have the problem to balance between "elaborate" test vs time. Would it make sense to have a set of tests that only run in the (I think weekly) master test?

In principle, yes, but this should not become an excuse for not writing more efficient tests.

[dweindl]

Does it make sense to perhaps test with specifically a model where it previously failed (e.g. Schwen_PONE2014 from #1465)?

Added a simpler test.

[PaulJonasJost]

Thanks :) If cloudpickle is more powerful then pickle, should we switch to cloudpickle everywhere?

[dweindl]

If cloudpickle is more powerful then pickle, should we switch to cloudpickle everywhere?

It seems pickle got the job done in most cases so far. I'd only do it where necessary.