Debugs for reading a .pickle file with signals#15
Debugs for reading a .pickle file with signals#15CodyCBakerPhD wants to merge 6 commits intoleiferlab:masterfrom
Conversation
CodyCBakerPhD
changed the title
| @@ -1,27 +1,6 @@ | |||
| #!/usr/bin/env python | |||
|
|
|||
| from distutils.core import setup, Extension | |||
There was a problem hiding this comment.
Of course, this PR solves the problem by completely removing the C++ stuff
If you'd rather keep it, there are some alternative approaches I can investigate (like making it optional - or splintering it to a different repo) but they will take more time
| import json | ||
| import pickle | ||
|
|
||
| import numpy as np |
There was a problem hiding this comment.
Moving this line was a minor style change - styling patterns such as isort recommend separating core Python libraries from third-party modules from the 'local' module
But if you prefer the first way just let me know; I didn't make this change holistically, but if you'd like that as well the easiest way would just be to setup pre-commit on the repo
| optogeneticsFrameCount = np.zeros(optogeneticsN, dtype=np.int) | ||
| optogeneticsNPulses = np.zeros(optogeneticsN, dtype=np.int) | ||
| optogeneticsRepRateDivider = np.zeros(optogeneticsN, dtype=np.int) | ||
| optogeneticsNTrains = np.zeros(optogeneticsN, dtype=np.int) | ||
| optogeneticsFrameCount = np.zeros(optogeneticsN, dtype=np.int64) | ||
| optogeneticsNPulses = np.zeros(optogeneticsN, dtype=np.int64) | ||
| optogeneticsRepRateDivider = np.zeros(optogeneticsN, dtype=np.int64) | ||
| optogeneticsNTrains = np.zeros(optogeneticsN, dtype=np.int64) |
There was a problem hiding this comment.
Numpy deprecated this behavior a while back (first a soft warning for several versions; then finally a hard removal that caused some of my errors)
There's actually 2 parts to it as well
i) np.int has become np.integer and is the 'base' data type of all specific np.int<number of bits> types
ii) they no longer allow dtype=np.integer which always defaulted a system specific np._int value - the modern recommendation is to set it to an explicit size
| import mistofrutta.struct.irrarray as irrarray | ||
| import wormdatamodel as wormdm | ||
| import savitzkygolay as sg |
There was a problem hiding this comment.
I moved each of these into the scopes of the functions that use them so someone can import the wormdatamodel package without needing to find each GitHub repo and do local installs
Using each function would still error if the package is missing, but at least that only occurs when (and 'if', in my case) those functions would be called
There was a problem hiding this comment.
An alternative to this, if you prefer to keep things as they are, might be to include git-based installations in the setup requirements for this package
That could however cascade a series of PRs like this one if there's any trouble installing those packages however
|
With the changes outlined on this PR I was able to successfully open a |
1 participant
When trying to open some
.picklefiles for the NWB conversion project, I had to make some changes to this repoWorking on Windows, Python 3.12, but a lot of this would affect other platforms and even older Python versions
matplotlibto setupI'd also recommend looking into getting this released on PyPI to simplify the installation process from requiring a git clone