
I feel sorry for anyone who legitimately needs to call in sick on GTA 6 release day.

Open source C and C++ software developer. miniaudio, dr_libs, and more. Discord: discord.gg/9vpqbjU

I feel sorry for anyone who legitimately needs to call in sick on GTA 6 release day.

I finally took the time to get this integrated into my development environment and it works well! Recommended. Is there anything like this for macOS and iOS? That would make my life so much easier.
Finally, no need to setup Windows just to test how my code compiles on MSVC.

GCC: note: '-Wmisleading-indentation' is disabled from this point onwards, since column-tracking was disabled due to the size of the code/headers miniaudio's power level is too high for GCC.

GitHub needs shared blocklists. For example, I should be able to set it up such that if @raysan5 blocks a user, that same user is automatically blocked for me too, and vice versa. We both had bad bug reports by the same user yesterday, all bulk created by AI.

miniaudio version 0.11.26 has been released. Release notes: github.com/mackron/miniaudio…

I've been maintaining some open source audio projects for 10+ years and never in that time did I bother to properly organise my test files. That's now been corrected: github.com/mackron/audiovect… I'm making this database public in case others might find it useful.

I'm using a library in C, and the maintainers have decided to omit parameter names in public function declarations. Honest question: How does this help the usability of your library? What has this actually achieved?

I've been experimenting with some AI assisted programming lately and overall it's quite good, but I've noticed that it can generate some horrendously inefficient code. It really likes doing short lived micro heap allocations. Once I even saw it do dlopen() on every key stroke!

I'm starting to understand the no-AI policy of some of the GitHub alternatives. For me, it's not about anti-AI ideology - I like AI - it's about having human communities and human connections, and not feeling like I'm on a dead platform full of bots.

I'm 90% confident that 90% of all PRs are now done by AI. Maintaining an open source project just isn't the same anymore. I think AI in programming is incredibly useful, but the feeling of less and less human interaction takes away a lot of the satisfaction of open source.

I've added support for proper gamma corrected text rendering, with optional LCD filtering, to my vector graphics library. Because heaven forbid anything on Linux gets gamma correction right...

So I'm working on a new single file vector graphics library. The software backend is starting to take shape! Still a lot to do.

I built a simple plain-text ticket system to help organise my work for my various projects. Each ticket is just a simple text file in a sub-folder. There's a VS Code extension for humans and a CLI for clankers to help work with it. github.com/mackron/ticketfil…

What's the biggest mistake you've made in software development in terms of maintenance costs? Mine: dr_wav's 2-pass metadata parsing system which I accepted from a PR. First pass calculates the required allocation size, the second pass fills out the data proper. Stupid mistake.

I think that would help solve some of the problems with slop PRs. Should be opt-in by project owners so that new contributors have a chance to start building up their respect rating. Is this a good idea, or is it stupid?

This is what I think GitHub needs: A "respect" system. PRs are rated "good" or "bad" by repository owners which affects the respect rating of the contributor. Then, project owners can set a threshold so that users with too low of a respect rating are banned from submitting a PR.