1.1.1

New features

Parallel reading of data

See test/testing.py for a full example. Modify the following around your APReader call:

import multiprocessing as mp
...


if __name__ == '__main__': # this line has to be included!
    # without 'processes=...'!
    pool = mp.Pool() 

    # pass the pool to the reader
    reader = APReader(file, parallelPool=pool)

    # make sure to close the pool after you are done with it
    mp.close()
    mp.join()

For the parallel loading to work, you have to define a parallel pool of processes in your top-level script. These processes will be accessed from within APReader-Functions. When passing no arguments to mp.Pool() it will automatically create as many processes as possible, according to the amount of threads your CPU allows (cores + virtual cores). It does not make sense to pass in more, since the APReader spawns the same amount of processes as there are CPU Threads. Increasing the amount of processes in your pool does not increase the amount of parallelism. It is fixed.

Keep in mind, that parallelisation is not always faster. Spawning of processes is expensive and can be wasteful for small files.

The results from APReader stay the same and you can continue your analysis.

Improvements

• Typo in Group.intervalstr fixed (micro and nano-seconds where swapped)
• Unit of Group.interval is now seconds

What's Changed

• Development on Version 1.1.1 by @leonbohmann in #18

Full Changelog: v1.1.0...v1.1.1