Tips#

git#

Update fork

git remote add upstream {url_to_remote}
git fetch upstream
git rebase upstream/main

Remove local and remote branch

Note that a common usage of a fork is to open a pull-request, in particular from a branch on the fork:

git checkout -b patch
... # commits here

Once merged, you can delete the branch:

git checkout main
git branch -D patch
git push --delete origin patch

Update remote branch

When a remote branch is deleted from a repository, it can still show up in the local branch list.

Often .. code-block:: bash

git fetch

should suffice. However, sometimes you’ll have to do

git fetch --prune

See this StackOverflow discussion.

Generate patch

In case a new release cannot be made (because you don’t have access to a repository), but you still want to update a conda-forge recipe, you can add ‘patches’ to the recipe.

To this end, clone the repository

git clone https://github.com/someuser/somerepo

Move to the repository

cd somerepo

and determine the number of commits since the last release (e.g. 3). Then have git create the patches, using:

git format-patch -3

This will create a number of *.patch files (3 in this case). These patches can be added to the recipe. See for example the cling-feedstock.

Partial clone

Recently I came across a few repositories with a very rich history (corresponding to a huge disk-space, requiring also a significant downloading time). If one is not particularly interested in the histories, on can partially (shallow) clone a repository:

git clone --depth=5 ...

conda#

Add library to Conda environment

Conda is a great tool as a package manager and virtual environment together. Many (up-to-date) packages are available through conda-forge. However, ever to often you might want to install from a local source (for example to test the current master branch).

Start by activating the relevant Conda environment (e.g. “myenv”):

conda activate myenv

(see Conda documentation on how to manage environments).

Then, go to the library:

cd /path/to/library

Python

Make sure that you have Python installed, e.g. using

conda install -c conda-forge python

Then, with the Python executable that is loaded (from myenv)

python -m pip install .

CMake

Make sure that you have CMake installed, e.g. using

conda install -c conda-forge cmake

Then, with the CMake executable that is loaded (from myenv)

cmake . -DCMAKE_INSTALL_PREFIX:PATH="${CONDA_PREFIX}"

Python#

Create movie with ‘ffmpeg-python’

ffmpeg-python is a great wrapper around ffmpeg to create your movie from Python.

Let us begin by setting up an environment that contains what we need:

conda activate myenv
conda install -c conda-forge ffmpeg-python
conda install -c conda-forge matplotlib

(see Conda documentation on how to manage environments).

Then, we will create an animation as a batch of images:

import matplotlib.pyplot as plt

filenames = []

for i in range(20):

    filename = 'image_{0:02d}.png'.format(i)
    filenames += [filename]

    fig, ax = plt.subplots()
    ax.plot([0, 1], [0, i])
    ax.set_ylim([0, 20])
    plt.savefig(filename)

To convert this to a movie, we will use ffmpeg-python:

import ffmpeg

(
    ffmpeg
    .input('image_%02d.png', framerate=2)
    .output('movie.mp4')
    .run()
)

pybind11 examples

pybind11 examples. Some basic examples on how to start using pybind11. Pybind11 is a C++ library that allows to expose a C++ library to Python easily.

ParaView#

ParaView examples

ParaView examples. Some basic examples on how to make data available in ParaView.