Closed kalenkovich closed 3 years ago
Hello @kalenkovich! Thanks for updating this PR. We checked the lines you've touched for PEP 8 issues, and found:
There are currently no PEP 8 issues detected in this Pull Request. Cheers! :beers:
There are some pep8 issues - I'll fix those before merging.
two things though:
contributing.md
. environment.yml
:
i'm not super familiar with it, but if i read the file correctly numpy
, sciypy
, and mne
are fetched twice - once via conda
in the environment.yml
file, and once via pip
in the requirements.txt
file. i assume conda will have priority. we could eliminate them from either of the requirement files Just complaining: The darndest thing is that it took me an extra million years just because
open
andcodecs.open
behave a bit differently on Windows. Because of that I always got an error when trying to deploy to PyPI. And the error message was so "informative" and the error itself was so common that I found this image long before I actually figured out how to fix it:
:D
- [x] setup works
- [x] coverage test works
- [x] tests passed
- [x] doc builds
Thanks for checking!
two things though:
- i'll make some inline suggestions to the instructions in
contributing.md
.
A bit confused about that one. You've already committed your suggestions, right?
- creating the conda environment from
environment.yml
: i'm not super familiar with it, but if i read the file correctlynumpy
,sciypy
, andmne
are fetched twice - once viaconda
in theenvironment.yml
file, and once viapip
in therequirements.txt
file. i assume conda will have priority. we could eliminate them from either of the requirement files
conda
won't automatically install stuff from requirements.txt
so there won't be any double installation. I am not entirely sure but I think some of the CIs might set up the environment using requirements.txt
so I'd rather keep it.
A bit confused about that one. You've already committed your suggestions, right?
yes, i was confused too :D my intention was to suggest changes, but then i did a normal commit.. can you remind me how to suggest changes via the github website? ppl in the mne prs do it all the time, but i didnt find it earlier.
conda
won't automatically install stuff fromrequirements.txt
so there won't be any double installation. I am not entirely sure but I think some of the CIs might set up the environment usingrequirements.txt
so I'd rather keep it.
alright, my bad.. i was reading to fast - i thought it would pip install the requirements.txt
in root as well
- pip:
- -r file:freqtag/tests/requirements.txt
- -r file:doc/requirements.txt
- twine
yes, i was confused too :D my intention was to suggest changes, but then i did a normal commit..
Ah. alright. My intentions and my results often diverge as well when it comes to GitHub :) My first ever PR: just merged it myself immediately after creating it. We then decided to change some settings so I wouldn't be able to merge at all :)
can you remind me how to suggest changes via the github website? ppl in the mne prs do it all the time, but i didnt find it earlier.
Sure! You go under the "Files changed" tab. There, you select a line or a couple of lines and press "+" to add a comment. In the comment toolbar above the editing area, there is a button with a plus-minus sign which allows you to add a suggestion alongside the comment.
Since otherwise everything seems to work, I'll just fix the PEP8 issues and merge.
I had to start from scracth. There is still no actual frequency-tagging code added yet. Just making sure everything works, tests are passed, docs are built, etc.
I did pretty much just what is described in the docs for the mne-tools project template here. Not everything worked straight away though - thus a thousand commits. Also, I am not sure what half the CI-like services are doing but I connected them anyway. Appveyor failing is fine, by the way: I haven't told it what to actually do yet.
@dominikwelke, could you please check that you can build the package, the docs and the tests are passed locally? I put the instructions under contributing.md.
Just complaining: The darndest thing is that it took me an extra million years just because
open
andcodecs.open
behave a bit differently on Windows. Because of that I always got an error when trying to deploy to PyPI. And the error message was so "informative" and the error itself was so common that I found this image long before I actually figured out how to fix it: