what happen to the nilmtk environment ?

[oneway3124]

Hi sir

conda create -n nilmtk-contrib-dlsc -c conda-forge -c nilmtk nilmtk-contrib

jupyter notebook

What happen?

But some days ago, it is not this state.

[PMeira]

Yikes! Looks like conda is buggy, as usual. It's still giving priority to old versions of the packages. For the time being, I just removed the old packages from the installation list to avoid this issue (they're now under the "retired" label, we can restore them later or the users can use the label explicitly if they really need old versions), but then it's installing very odd versions for some of the dependencies.

Please note that, from #52, it might be possible that the latest TensorFlow (2.4) breaks some methods, so it's a good idea to pin it to a lower version (try 2.3 or 2.2) (EDIT: your original command seems to be working now.)

We'll release an updated package soon (few days) to handle that.

Given this weird state of things, if you still get a broken installation, please try installing mamba -- see https://github.com/mamba-org/mamba

If you use mamba instead of conda in your original command, it was picking the right versions for me. I guess we should update the docs to recommend it instead of conda. It's also much faster. There is a distribution with it preinstalled: https://github.com/conda-forge/miniforge#mambaforge

[oneway3124]

I have tried to reinstall the conda environment completely. Wait for some new message.

[oneway3124]

@PMeira Can you help me about this reinstall?

[oneway3124]

This error is still . AttributeError: 'str' object has no attribute 'decode'

[om-maheshwari123]

@oneway3124 how did you tackle AttributeError: 'str' object has no attribute 'decode'?

[ArunRanganarsimhaiah]

@oneway3124 how did you tackle AttributeError: 'str' object has no attribute 'decode'?

conda install -c anaconda h5py=2.10