Git issues I love/learned a lot from!
One highly underrated thing about open source scientific software is the scientific discussions that happen on forums, issue trackers, or a random web link with notes on how to do something niche with a program-of-interest. Quantum chemistry program exchange (QCPE) is one of the earliest forums of source code sharing, and we are in the age of githubs and gitlabs. The revered tradition of discussing computational chemistry on a mailing list, such as CCL.net , is slowly transforming into a discussion on git-hub(lab) issues. Here are some of my favourite issues (in other words bookmarked for posterity) within the programs and software ecosystem that I use the most, or I interact the most with (in no particular order):
Stereochemistry in small molecules and how they are handled by various cheminformatics toolkits (OpenEye and RDKit listed here). This one dates back to the inception days of openforcefield and one of the long standing issues on the openff-toolkit tracker and a great reference - https://github.com/openforcefield/openff-toolkit/issues/146.
A discussion on my dearest quantum chemistry engine, Psi4, about a smart SCF solver - https://github.com/psi4/psi4/issues/888. I like the evolution and implementation of ideas into code, figuring out tough edge cases to validate and perform a quality check, and lot more lessons from this issue.
A complete list of how the central atom in an improper torsion definition is indexed in different Molecular Dynamics programs - https://github.com/lilyminium/improper_orders. The inconsistency in the specification is one of many things a developer has to be extremely careful about.
Energy decomposition in OpenMM, how different force groups in a simulation can be labeled - https://github.com/openmm/openmm/issues/1463. This is extremely useful when introducing custom forces, or if we just want to take a look at the contributions coming from bonded and nonbonded interactions.
A list of numerically unstable functionals as seen on the LibXC tracker - https://gitlab.com/libxc/libxc/-/issues/400. It includes a link to a preprint as well.
A comprehensive list of outstanding issues, feature requests, and ongoing development activities related to free energy calculations in GROMACS, a widely used MD engine - https://gitlab.com/gromacs/gromacs/-/issues/3945.
I could go on listing a lot of these, but the main point is that structured discussions like these are at the heart of open source science and help with knowledge transfer to the next generation. I would love to see other git issues you learned the most from, or you refer to pretty often.
Ā