Dependency on numpy version

[krisvanneste]

Dear Albert,

In your recent commit "Add optimized approach", you replaced the "interpolation" argument of the numpy percentile function with the new "method" argument, which was introduced in numpy 1.22.0. Would you consider making pyrotd independent of the numpy version, by either:

• removing the interpolation/method argument, as "linear" was and still is the default value
• checking for the numpy version, and pass the correct argument, something like:

  if [int(val) for val in np.__version__.split('.')] < [1, 22, 0]:
  p_peak_resps = np.percentile(rotated.peak_resp, percentiles, interpolation="linear")
  else:
  p_peak_resps = np.percentile(rotated.peak_resp, percentiles, method="linear")

  Unfortunately, I depend on a slightly older numpy version (1.20.2) in my environment...

[krisvanneste]

Thank you!

[arkottke]

No problem. Thanks for pointing out the problem. I ended up using: https://numpy.org/doc/stable/reference/generated/numpy.lib.NumpyVersion.html to test the version.

On Fri, Jun 30, 2023 at 9:45 AM krisvanneste @.***> wrote:

Thank you!

— Reply to this email directly, view it on GitHub https://github.com/arkottke/pyrotd/issues/7#issuecomment-1614912174, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAH4LMY7RMS3NPQ2FET5LCTXN37BVANCNFSM6AAAAAAZUCSN7A . You are receiving this because you modified the open/close state.Message ID: @.***>

[krisvanneste]

I will test it in the coming days.

[krisvanneste]

It works!