Conda failing to resolve when pymapd is in dependencies

[datapythonista]

Not sure the exact reason, but when running conda install -c conda-forge pymapd, the conda solver takes a very long time.

And if pymapd is in a requirements file with other dependencies, the resolution may not be feasible (the solver takes more than one hour, or never finishes).

I guess there is something in pymapd dependencies that is causing this. It's happening at least with Python 3.6 and 3.7.

[xmnlab]

hey @datapythonista could you provide the environment pkg list you used to install pymapd?

in a new conda environment, the command conda install pymapd took ~22 s

real    0m22,116s
user    0m18,043s
sys     0m1,888s

[datapythonista]

I thought I tried it in a clean miniconda environment, but apparently now. I tried now, and it resolves in a reasonable time.

Not sure what was the dependency or dependencies that were installed that were causing the problem. I'll leave the issue open, since I still think there is a problem with the dependencies. I'll try to have a look to see if I can find the package that is making the solver be super slow when solved together with pymapd.

Thanks @xmnlab

[vaceslav]

Is it possible to install pymapd with latest version of rapids 0.17 and cudatoolkit 11.0? I can't install it in a new environment.

Only rapids 0.14 and cudatoolkit 10.2 is possible.