Initial release of lava-dnf v0.1.0

[mathisrichter]

Starting the code review process of the initial release of lava-dnf v0.1.0.
Since there is a lot of code, I am doing this now already.
The following things still need to be changed - which we will hopefully be able to do tomorrow:

• Plotting needs to be integrated. This will be based on the Monitor feature that is awaiting code review in lava-nc/lava.
• The tutorial has to be finished with inputs and plots.
• Multiple unit tests are currently skipped because they have non-deterministic behavior. We are looking into that.
• init.py files have to be changed to the new convention.
• Unit tests are currently failing because requirements.txt is depending on the last Lava release 0.1.1 rather than the current version in the repository.
• All comments start with non-capitalized letters - I will change that, no need to remark on every instance.

The modules that need the most critical review are:

• lava/lib/dnf/connect (a function that connects two (neuron) processes)
• lava/lib/dnf/inputs (runtime-adaptable spike generator process)
• lava/lib/dnf/kernels (two kernels that can be used by a convolution-like operation)
• lava/lib/dnf/operations (operations that are used to parameterize the connect() function)

I suggest you pick one of these modules and also go through its unit tests rather than trying to cover all the code.
If you want to get an overview of how the user is supposed to use the whole thing, have a look at the tutorial (lava-dnf/tutorials/lava/lib/dnf/tutorial_dnf_101.ipynb) and the acceptance tests (lava-dnf/tests/lava/lib/dnf/acceptance).

[bamsumit]

Didn't go through all files. Looks really good @lava-dnf team.

[mathisrichter]

Merge and unit tests depend on PR#96 and changes to Dense in lava-nc/lava.

[awintel]

I have not really had a chance to review the code much. It all looks really nice and polished.
However, I noticed a lot of " -> None" for functions that return nothing.
That is not wrong but also seems pretty unnecessary. Just add type annotations when there is a type.

[mathisrichter]

I have not really had a chance to review the code much. It all looks really nice and polished. However, I noticed a lot of " -> None" for functions that return nothing. That is not wrong but also seems pretty unnecessary. Just add type annotations when there is a type.

I am adhering to the flake8 type annotations checks. I also feel that the None return types are a bit overkill. When we have some time to add type annotations everywhere in Lava, we should agree on which checks we want globally and which to ignore. At the moment I am using the ones @bamsumit suggested, assuming he is using the same for lava-dl.

pip install flake8-annotations
flake8 --ignore=ANN101,W503,E501 src/lava

Although ANN101 (type annotations for function arguments) and E501 (line length) seem important to have.
https://www.flake8rules.com/