Guitar hero inspired me to pick up the guitar as a kid. My dad had tried and failed for years

lmao I just checked all the new features. I'm psyched for a working pause

arc witch. might try ranger or templar next. have you been playing long?

Defs replace as much of the standard libraries as you can. Plenty of great stuff on GitHub.

All I use is:
Malloc/calloc/realloc/free
Memset/move/cpy/cmp
snprintf
math.h

All other string and i/o functions are awful.

Have you considered switching to plain old C, or something even newer and riskier like Zig, Odin or Jai?

Good design makes gotchas like this hard to make, and I certainly haven’t found the right one yet. If you have ideas or tips to share, I’d love to hear it!

Also if you hit me up on discord or something I’m happy to share code

Extra logic is required on mouse exit events to stop this trip up happening. I suspect I’ll run into similar problems with keyboard focus gained lost events, and maybe drop + drag end events.

A gotcha I ran into with this no callback approach is responding to mouse exit/enter in the correct order. For example, two widgets A & B set/restore a cursor on enter/exit events. If A is drawn before B and the user moves the cursor from B to A, then B’s exit event will be resolved last.

Yep, you basically described the code I have so far (ids and bounds checking).

Currently z indexing is an unsolved issue. I’m hoping clever use of a “generational ID” will help. That’s a trick game developers use.

Definitely. The latency thing is a year old issue on GitHub, and probably very easy to solve, I just haven’t gotten around it lol

I’m struggling to understand your reply. Are you saying certain widgets require callbacks?

If so, I believe key/mouse events can be stored in a queue or similar, then consumed at the site of the widget without any requirement to register a callback.

Good luck! Are you using another lib, or writing your own?

You’re not missing out

Removing a retained mode widget is just as bad. First you remove it from the header, then remove all references to it, possibly across several files and definitely in several functions/methods.

What a nightmare!

This developer friction certainly stops many great ideas from being tested!

The equivalent of this is extremely cumbersome in a retained mode GUI (eg. JUCE, VSTGUI, Qt). To add a widget you will probably add a new member to a class in a header file, then it needs to be initialised correctly in a constructor within a .cpp file, then it needs to drawn in some third place...

If you want to test a GUI idea using immediate mode widgets, it's as simple as writing all necessary code in one place. If you need to temporarily make space for something, then removing widgets is as easy as commenting out a block of code, or a function call.

In C/C++, you can literally plot any chart to see and hear results in real-time. You can add/remove temporary "static" parameters for testing (see above pic) and commit them to proper plugin params later

There is also no algorithm rewrite from mathematical notation to C/C++