Quantum-Dynamics-Hub / libra-code

https://quantum-dynamics-hub.github.io/libra/index.html
GNU General Public License v3.0
50 stars 46 forks source link

Actual *installation* instructions ? #122

Closed mboisson closed 3 years ago

mboisson commented 3 years ago

Is there a way to actually install this software ? Building and then editing .bashrc files and setting path is hardly an installation. https://quantum-dynamics-hub.github.io/libra/installation.html

How about a "make install" target ? or a way to build a python wheel that can then be installed with pip ?

I tried

[mboisson@build-node build]$ make install
make: *** No rule to make target 'install'.  Stop

I tried

[mboisson@build-node libra-code-5.0.0]$ python setup.py bdist_wheel
running bdist_wheel
running build
running build_py
running egg_info
writing Libra.egg-info/PKG-INFO
writing dependency_links to Libra.egg-info/dependency_links.txt
writing top-level names to Libra.egg-info/top_level.txt
reading manifest file 'Libra.egg-info/SOURCES.txt'
writing manifest file 'Libra.egg-info/SOURCES.txt'
UPDATING build/lib/src/_version.py
error: [Errno 2] No such file or directory: 'build/lib/src/_version.py'
alexvakimov commented 3 years ago

Hi Maxime,

First of all, this path https://quantum-dynamics-hub.github.io/libra/installation.html is a bit outdated. The latest one is the README on the devel branch. Please try that one - this may actually help you solve the other issue you mentioned earlier today. The most difficult part of installing/building it is the creation of the correct Conda environment.

As for the actual install question - I think, this could be done at some point, but I never had a chance/priority to dig into it. If you have an idea for how this could be added - I'd be happy to consider your pull request (giving your the proper credit, of course).

Also, if you like - we can discuss more on the Slack channel: https://join.slack.com/t/quantumdynamicshub/shared_invite/zt-mjbhjssx-GGhsbYHxeBMvhmumK_j7LA (this should be a permanent link)

mboisson commented 3 years ago

I'll have a look at the README.

For what it's worth, conda is completely out of the question. This is for a HPC cluster, and conda does not play well with clusters (https://docs.computecanada.ca/wiki/Anaconda/en)

We already have the dependencies, and we are not reinstalling them with conda.

From a software packager point of view, what is needed is : 1) what are the dependencies 2) how to install your software, once those dependencies are installed

installing the dependencies should not be part of the installation procedure, and there should not be an assumption made that users will install them with conda.

alexvakimov commented 3 years ago

Yes, whatever works