scratcharchive / kachery Goto Github PK

Hi @magland

I'm trying to use this package standalone from spikeforest. I think I followed the install directions correctly, but I'm running into an error when trying to get things from remote locations. For example, when I run the test_remote.py script in the repo, I get this error:

Exception: No password found for server.channel: default_readwrite.neuro

Ultimately, I'm trying to get the recordings from spike forest (which fails for me too)

It seems like there's a setup issue on my end, but I couldn't find in the documentation anything about this. Apologies in advance if I overlooked something.

-John

For example, if we do ka.load_file('sha1://....', fr='default_readonly')

and the daily quota for download on this channel has been exceeded, then we need to provide a helpful error message rather than just a server error.

PIP installed kachery into a clean venv environment, attempted to run sample code, and encountered the error below. Resolved the problem by installing numpy. Recommend adding numpy to install_requires section of setup.py.

(kachery) ~ $ kachery-store test.py Traceback (most recent call last): File "/home/rogerhurwitz/venv/kachery/bin/kachery-store", line 7, in <module> import kachery as ka File "/home/rogerhurwitz/venv/kachery/lib/python3.8/site-packages/kachery/__init__.py", line 1, in <module> from .core import set_config, get_config File "/home/rogerhurwitz/venv/kachery/lib/python3.8/site-packages/kachery/core.py", line 2, in <module> import numpy as np ModuleNotFoundError: No module named 'numpy'

I recently became acquainted with the Click python command line interface library, and thought that it might be a good fit for kachery. It looks like you are writing a lot of "glue" code to compensate for argarse's shortcomings that Click gives you for free (e.g., type checking, environment variable parsing, sub commands, etc.). For illustration purposes, I mocked up the attached examples.

kachcli.zip

Hi !
I have been trying to run spyKingCircus through the docker wrapper.
Initally I did not face any issues with running it in the container.
But recently I have been facing this issue
Traceback (most recent call last):
File "C:\Users\Maitreyi\CareerDevelopmentoversumeer\SummerResearch2020\spikeforest2\kachery\example_mountainsort4.py", line 12, in
result = sorters.mountainsort4.run(
File "C:\Users\Maitreyi\CareerDevelopmentoversumeer\SummerResearch2020\spikeforest2\kachery\hither_sf_core.py", line 271, in run
_run_job(job)
File "C:\Users\Maitreyi\CareerDevelopmentoversumeer\SummerResearch2020\spikeforest2\kachery\hither_sf_core.py", line 449, in _run_job
returnval, runtime_info = run_function_in_container(
File "C:\Users\Maitreyi\CareerDevelopmentoversumeer\SummerResearch2020\spikeforest2\kachery\hither_sf_run_function_in_container.py", line 263, in run_function_in_container
raise Exception('Non-zero exit code ({}) running [{}] in container {}'.format(retcode, label, container))
Exception: Non-zero exit code (127) running [mountainsort4] in container docker://magland/sf-mountainsort4:0.3.2

I am trying to benchmark my algorithm with all these standardized algorithms and I am finding spike forest super helpful, but I dont know why I am suddenly having this error.

would it make sense to include a way to remove files? perhaps something like ka.remove(some_hash).

alternatively, there could be a garbage collection mechanism.

Hi!
I have been attempting to use spike forest for some spike sorting benchmarking .
I was able to clone the repo and I also downloaded the dependency Kachery-but I am having the following error when I attempt to run one of the examples in this repo(running spykingCircus sorter)

with ka.config(fr='default_readonly'): (Error in this line)
AttributeError: module 'kachery' has no attribute 'config'

I have tried to set the environment variable -kachery storage to some folder for temp files storage and I am not too sure how to set the password for the environment variable kachery_password.

I also attempted to run the test files in the kachery repo and I get the error-Permission denied.

I am suspecting that there is some issue with the kachery dependency.
Any help/instructions will be highly helpful.

Regards
Maitreyi

Hi Jeremy,
I like most of the documentation, but think it needs to be expanded in 2 ways:

1. including more example use-cases up front in the intro. Why do people/labs need this?
2. Tutorial needs to include the basic set-up for local directory. Eg when I install and run
   kachery-store on a directory, it says
   WARNING: please use the KACHERY_STORAGE_DIR environment variable.
   and apparently nothing is stored (this feels more like an ERROR than a warning).
   I did
   export KACHERY_STORAGE_DIR=/tmp/
   and then it works and creates /tmp/sha1/ and /tmp/sha1-cache/
   I think you need to guide the user through that process, encourage them to look at the result.

Only then move on to the remote server aspect. The docker-instructions for remote server is a little loose too - what is magland/kachery2 supposed to be for the user? An example set-up is needed here, maybe some flow diagrams.
Best, Alex