Jupyter

[felipeZ]

Create jupyter notebooks for the tutorials:

• Gromacs QMMM
• LAMMPS QMMM
• DFT-GWBSE
• KMC

[ghost]

Congratulations 🎉. DeepCode analyzed your code in 0.365 seconds and we found no issues. Enjoy a moment of no bugs ☀️.

👉 View analysis in DeepCode’s Dashboard

Edit DeepCode’s Configurations here

---

[felipeZ]

@JensWehner the Notebooks are ready, they take considerable time to run and therefore running then in the CI is not feasible, what shall I do then?

[felipeZ]

@JensWehner @junghans How can I know if the Jupyter notebooks are running correctly?

[felipeZ]

@JensWehner @junghans How can I know if the Jupyter notebooks are running correctly?

[felipeZ]

I just got confused with the CI. Now I see it

[felipeZ]

@junghans can I skip the Sphinx building in ubuntu? It is impossible to generate the documentation from the jupyter notebooks if there is not nbconvert available

[junghans]

I think the previous solution was better now the build requires nbconvert even to build xtp without any documentation.

[felipeZ]

I have made a revert. The problem is that sphinx needs nbconvert to create the documentation and there is not suitable nbconvert for ubuntu. So, can we skip the Sphinx build on ubuntu

[JensWehner]

We only have to use it on one single platform. Despite all that, @felipeZ could you add sth. to the developer guide in votca/votca about what the documentation does?

[junghans]

I have made a revert. The problem is that sphinx needs nbconvert to create the documentation and there is not suitable nbconvert for ubuntu. So, can we skip the Sphinx build on ubuntu

Wouldn't

find_package(JUPYTER_NBCONVERT 5.4.1)
if(JUPYTER_NBCONVERT_FOUND AND SPHINX_FOUND)

take care of that?

[junghans]

We only have to use it on one single platform. Despite all that, @felipeZ could you add sth. to the developer guide in votca/votca about what the documentation does?

Correct. We can exclude stuff in the setup action, but I don't want the user to not being able to build xtp on Ubuntu because JUPYTER_NBCONVERT is required. We should try to find a new enough version and if that fails, we just skip that part of the documentation build.

[felipeZ]

The problem is the following, there is a index.rst file in VOTCA/VOTCA that contains a reference to the XTP-TUTORIAL.rst file. When sphinx is trying to build the documentation it will go first to index.rst then to XTP-TUTORIAL.rst. The problem is that inside that last file there are reference to the jupyter notebooks and sphinx will always try to convert those Jupyter notebooks into documentation by running them.
On the other hand, the statement:

find_package(JUPYTER_NBCONVERT 5.4.1)
if(JUPYTER_NBCONVERT_FOUND AND SPHINX_FOUND)

Can just try to run the notebooks or just copy them as they are, but Sphinx will always try to run them anyway and seems there is no nbconvert tool it would fail.
So, I agree with @junghans that the best alternative is just excluding the Sphinx build on Ubuntu.

[junghans]

If we exclude it in the CI it will still trip up user on ubuntu running it manually.

Can we tell Sphinx to ignore the broken links? Or just create a mock version of xtp-tutorials.rst?

[felipeZ]

We can ignore broken links, with nbsphinx_allow_errors = True in the config.py file. I am going to do a PR

[junghans]

On openSUSE: ModuleNotFoundError: No module named 'pandas'

On Fedora Rawhide: ModuleNotFoundError: No module named 'seaborn'

[junghans]

On openSUSE:

Notebook error:
PandocMissing in DFTGWBSE_ENERGY.ipynb:
Pandoc wasn't found.
Please check that pandoc is installed:
http://pandoc.org/installing.html

[felipeZ]

@junghans @JensWehner It is working!!

[JensWehner]

@felipeZ good work.

[junghans]

Fixup in #26