Best practices in scientific software development

[Beliavsky]

I found the analogies between C++ and Fortran especially interesting. Geert Jan Bex co-authored a Fortran MOOC and the funny video where Fortran devours a Python. A Fortran-specific version of this material could be created. For example, Jan Bex recommends CMake for builds, but a Fortran course would highlight FPM.

---

Material [by Geert Jan Bex] for a training on best practices for programming and software development.

Programming languages

Although this training aims to be programming language-agnostic, the repository also list a number of tools that are programming language-specific. Obviously, this can not be exhaustive, so feel free to suggest additional tools if you are aware of any.

Programming languages covered:

• C
• C++
• Fortran
• Python
• R

Table of contents

1. Syntax versus semantics
2. Code style and conventions
3. Version control & collaboration
4. Code documentation
5. Testing

• Testing as experiments
• Unit testing
• Functional testing
• Code coverage

6. Deployment
7. Continuous integration
8. References

[hkvzjal]

There is an ongoing initiative in this direction centered around Fortran here: Course Curriculum Outline · carpentries-incubator/intro-to-modern-fortran · Discussion #6 · GitHub

The program you outlined is of course way more general! they are quite complementary I think

[nbehrnd]

Let’s invite Geert Jan Bex with @gjbex to join the thread.

[gjbex]

To put this into context: I was asked to provide a 3-hour best practices training that would be high-level and programming language-agnostic. That training is tomorrow, and I created this website to allow the participants to revisit the topics later.

Many software engineering concerns are indeed not specific to a particular programming language, but tools and many aspects of code style are.

Some of the material was taken from an MOOC on defensive programming and debugging that I produced a couple of years ago and that unfortunately is no longer online. Other material ties in with many of the programming language-specific training sessions I do.

My intention for now is to add to this repository/website when I come across interesting tools or material. We’ll see where it goes.

[nbehrnd]

@gjbex

Some of the material was taken from an MOOC on defensive programming and debugging that I produced a couple of years ago and that unfortunately is no longer online.

I guess it is the one ([once] equally) on futurelearn here with a companion page still accessible here.

[nbehrnd]

Three resources I’m aware of which seem to resonate with this thread are

• the BCI Research Software Engineering Guide by the Bristol Composites Institute, University of Bristol (last edit by 2022)
• a compilation by West Virginia University about scientific programming, sub sections straddle Fortran, C, C++; Python; R; Julia; Matlab; Perl; OpenMP, MPI, OpenAcc, Cuda (last edit by 2023)
• the presentation Best Practices for Programming GPUs using Fortran, OpenACC, and CUDA by David Appelhans (Senior DevTech @NVIDIA, March 2023) – though frankly (because I don’t use GPUs) this is a resource I only know from the outside/by name thanks to an exchange around r/scientificcomputing.