Ask HN: What's your experience working on software for science?

4 points·temporalparts··5 comments

Hi HN!

I'm giving an invited talk at APS March Meeting next week about the importance of software for progressing science. I have deep personal experiences, but I'd love to hear from this community!

Have you worked on scientific software? What was your experience? What went well? What went sideways? What advice do you have for others trying to make an impact?

View on HN

I've done a lot of engineering software, and some software for science. I work with a lot of people in this space. On a whole I find it interesting that the defacto platform for these pursuits seems to be shifting away from MatLab and towards Python.

On a whole I think this is a good thing as there's less vendor lock-in and ultimately a smaller gap between the science software being developed and consumer/end-user software. This has also proven to help with collaboration between research and industry.

I worked on lab automation and scientific software early in my career. One piece of advice: keep it ruthlessly simple. In most software a bug is a bad UX. In scientific software a bug can mean contaminated samples or months of invalidated research. Every unnecessary abstraction is another place for something to go wrong silently.

I have, a lot. Advice: "Actually you don't have time to not write some sodding tests"

Would love to learn more! What did you work on (even vertical: chemistry, medicine, materials, physics, cosmology, etc.), notable stories?

Can't really put my CV here, but a dozen years as Applied Maths postdoc in remote sensing and space applications. A lot of reimplementation of F77 and Matlab research code to C/C++ and Python with some sodding tests.

Reply on news.ycombinator.com