-
Notifications
You must be signed in to change notification settings - Fork 86
Anaconda deployment #52
Comments
For the record, installations without mkl in conda seem to work out of the box, while MKL support within conda requires further work as neither compilervars.sh nor mklvars.sh is available in conda MKL. I had no luck specifying the paths alone:
This is a working example for the openblas route (note the nomkl package)
|
The nomkl was not working for me, so I installed the python from conda-forge: However, qml depends on ase, but it wasn't installed on the new environment as it's not in the requirements list, so I had to: After that, the package worked like a charm. |
Hi!
|
@mrks26: guessing from the output, it seems as if the detection of the location of one dependency (MKL) did not succeed. As far as I know, qmlcode is not regularly tested for use under Windows. Personally, I use conda within "Windows Subsystem for Linux (WSL)" which is nicely integrated and simple to install via the Microsoft App Store. Effectively it is a Linux VM. qmlcode works there out-of-the-box. Perhaps a workaround that does not take too much time. |
Thank you very much for your quick reply and the helpful suggestions! |
Build anaconda package and deploy it to anaconda cloud in travis. Docs on how to do this: https://conda.io/docs/user-guide/tutorials/build-pkgs-skeleton.html
The text was updated successfully, but these errors were encountered: