I’m not complaining, but I didn’t realize how much work it was. It makes me really respect the people who do it on a regular basis.
For example:
- You know how to use your software, but other people don’t. So you need to write documentation.
- You can just modify the source files, but it’s impractical for everyone to do that. So you need to add a config file.
- You can just drag the output files into place, but that’s impractical for everyone to do. So you need to package it.
- You trust yourself, but distro maintainers rightfully don’t. So you need to package your source code and configure the package to compile it.
- You will abide by your idea of how the software should be used, but other people might not. So you need to pick a license.
Sometimes I think there must be an easier way, but I can’t think of any. I guess it probably gets easier with experience.
That’s why engineers are, on average, paid more than researchers… And why research is such a nicer job.
Create robust and easy-to-use stuff is tough and you don’t get much reward
But if anything goes wrong or doesn’t work right, suddenly the users remember who deserves recognition
I’ve been attempting to build systems to make this “robustness” redundant across all my works, but I always feel there’s something more that I missed. I can’t tell if this task is simply never-ending or I just lack the knowledge of covering all the dots from the get-go or both.
I’ve seen a chart where software development complexity grows 3 times on each of these steps. For example:
- implement a business logic: 1 day
- cover with tests: 1 * 3 = 3 days
- write documentation: 1 * 3 * 3 = 9 days
- make it installable by other users: 1 * 3 * 3 * 3 = 27 days
This is basically the problem with “suckless” software summed up.
Even the step from one-off script to reusable code is a 1:3–10 rise in complexity, even without distribution. It’s really interesting how problems grow.
Documentation is very useful today (to clarify our thoughts on what is useful and what is not, what is in scope and what is not), and for our future selves.
Writing small bits of software made me appreciative of the work teams put on large pieces of infrastructure!
Hard agree on helping out your future self. I routinely drop a commands.md file in every project now, and dump any commands in there for creating the dev environment, the build step, any thoughts that might help when I come back in five years.
