cubicweb-timeseries #3851117 consider changing db serialisation of timeseries [open]

see:

(12:49:56) alain: timeit.timeit(stmt='import numpy,cStringIO; numpy.savez(cStringIO.StringIO(), numpy.random.random(1000000));', number=10)
(12:50:02) alain: 0.747920036315918
(12:50:15) alain: timeit.timeit(stmt='import cPickle, numpy, zlib; zlib.compress(cPickle.dumps(numpy.random.random(1000000)))', number=10)
(12:50:20) alain: 46.006000995635986

That would entail a big nasty migration, but so what (?).

prioritynormal
typeenhancement
done in2.0.0
closed by<not specified>