Closed bernt-matthias closed 1 year ago
test fals currently because requirements.txt will be added only in https://github.com/yigbt/deepFPlearn/pull/11
Looks good to me. There is one thing: Checking the order of imports with isort
lets the action fail. I have no strong preference, but usually I import system libs first and references into our own package after that. If you also tend towards this, then you need to get rid of the isort
checking. Otherwise, the imports where the "check sort action" breaks need to be fixed.
I think it's ready to be merged after that. Sorry, for taking so long. "Someone" had to remind me.
There is one thing: Checking the order of imports with isort lets the action fail. I have no strong preference, but usually I import system libs first and references into our own package after that. If you also tend towards this, then you need to get rid of the isort checking.
This is fine .. isort
does exactly this.
Now all works, except for typing :( Seems not trivial to fix. Should we:
1) skip for now and open a separate PR what adds and fixes typing 2) fix it here?
Maybe this is a stupid question, but why is mypy
not in the requirements.txt
if you want to use it? As far as I can see from the GH Actions log, this here
indicates that it's just not installed. Am I missing something?
Maybe this is a stupid question
There is no such thing as a stupid question :)
The requirements.txt
doesn't seem to work yet, and I had to try it locally with the Python version GH action uses. The package data-science-stubs
is nowhere to be found, and I replaced it with data-science-types
. Also, flake8
is unavailable in version 6.0.0 for this Python version. What worked is
black==23.3.0
data-science-types
flake8
isort==5.12.0
mypy==1.2.0
pip==22.0.4
pytest==7.1.1
types_setuptools
types-tensorflow
typing-extensions
-e .
However, running mypy
results in 93 errors. So unless @mai00fti is willing to tidy up the code and make necessary changes, I recommend skipping the mypy
test.
Hi @mai00fti can you check the CLI tests. Learning seems to be to fast (compared to the runtime expectations that you mentioned), ie all three calls need 2:30min.
For prediction
I see two warnings:
WARNING:tensorflow:No training configuration found in save file, so the model was *not* compiled. Compile it manually.
WARNING No training configuration found in save file, so the model was *not* compiled. Compile it manually.
Can they be ignored?
Also convert tells me that there is nothing to convert...
INFO Convert all data files in /home/runner/work/deepFPlearn/deepFPlearn/example/data
INFO Found 0 files to convert
So for each PR the workflow will check python linting and pytests is successful.