Closed h-mayorquin closed 2 years ago
Never used the make_env.yml myself - go right ahead if you want to synchronize the two.
as I have been using it in dandi to run our conversion.
Not sure what you mean by this, though - you should just be able to clone the repo, pip install -e buzsaki-lab-to-nwb
in a fresh environment, and run it from there. Full environment specification is more for communicating python versions and/or conda vs. pip install sources
pip install -e buzsaki-lab-to-nwb
is failing on dandy for me and producing the following error:
ERROR: No matching distribution found for scipy==1.4.1
The following just worked fine for me on the Hub
# log in
bash
conda create -n test_buz_install python=3.7
git clone https://github.com/catalystneuro/buzsaki-lab-to-nwb
pip install -e buzsaki-lab-to-nwb
Do make sure you always have the intended python version and a fresh environment.
In debugging matching distribution errors, it's also helpful to check PyPI itself for published versioning: https://pypi.org/project/scipy/#history
Also, if all else fails in these situations, you can always precisely match the installation structure of the continuous integration test that routinely verifies the package is installable in this fashion - https://github.com/catalystneuro/buzsaki-lab-to-nwb/blob/master/.github/workflows/install.yml
@CodyCBakerPhD can we bump up the scipy dep version? I don't think it is compatible with the latest versions of Python
@bendichter Sure, I can just update everything to 3.9 and include looser versioning on the central tools as well. I'll try to get that out tomorrow
@CodyCBakerPhD See my PR here
The following just worked fine for me on the Hub
# log in bash conda create -n test_buz_install python=3.7 git clone https://github.com/catalystneuro/buzsaki-lab-to-nwb pip install -e buzsaki-lab-to-nwb
Do make sure you always have the intended python version and a fresh environment.
In debugging matching distribution errors, it's also helpful to check PyPI itself for published versioning: https://pypi.org/project/scipy/#history
Indeed, as per Ben's comment, it was the version of Python causing the problem.
Question, the latest version of nwb-conversio-tools
is one patch ahead of our requirements here (requirements here are nwb-conversion-tools==0.9.3
whereas the latest version is nwb-conversion-tools==0.9.4
. How do we handle these issues? when do we align the requirements? with minor releases?
Question, the latest version of nwb-conversio-tools is one patch ahead of our requirements here (requirements here are nwb-conversion-tools==0.9.3 whereas the latest version is nwb-conversion-tools==0.9.4. How do we handle these issues? when do we align the requirements? with minor releases?
There shouldn't be any major breaks to back compatibility until major version increments like v0.10.0
, so in that case we keep versions in a repo like this pinned to whatever. I guess we really should just make different requirements for each conversion though unless we do want to go back and update previous conversions with new API changes. For all intents and purposes, you can always just use master branch of nwbct and you'll be fine.
@h-mayorquin Can you make a PR for this? Should be really easy, I know you were kind of already doing this locally on your end.
The contents of the file
make_env.yml
seem to be outdated and out of synch withrequiremens.txt
:make_env.yml
https://github.com/catalystneuro/buzsaki-lab-to-nwb/blob/7e7c9c2f7469c652b3de3f1d6bc2ffe06b24d426/make_env.yml#L1-L21requirements.txt
https://github.com/catalystneuro/buzsaki-lab-to-nwb/blob/7e7c9c2f7469c652b3de3f1d6bc2ffe06b24d426/requirements.txt#L1-L19Is there a reason for this mismatch? I am intending to update the
make_env.yml
file to make it match as I have been using it in dandi to run our conversion.Can I move forward with this @CodyCBakerPhD ?