Thank you for trying out game and writing about it! Sorry we missed you at the booth. We'll def send a steam key your way when we'll have a more mature game in our hands :)

Example of what a bot farm looks like for people wondering.

still messin' with erik (synth prototype). random technical thread, so i'll open with a vid: as an experiment, all code runs once per '4 samples', at 16khz; the only data type is a float4 ie 4 samples in a row, or a single note: pitch, gate, sequence and sample no (think: tracker cell)

When your gamedev friend shows you progress on their new engine.

Damn, the save upgrade thing must have been painful. Thanks for the link, the presentation is quite interesting

Now I have to prototype it quickly, and hope I don't find a flow in my solution. That would be a feeling of the same magnitude, in the other direction!

Just thought of a GREAT solution to a game engine design problem that was rolling in my head for a couple months. What an amazing feeling when it happens!

I'm honestly fine with explicitly passing an ID for the "yield" locations. That way when modifying the code, it's still possible to remain compatible with an old save by thinking through each possible resume location. Interesting idea to do a preprocessing step, some sugar would be nice indeed!

interesting. Did you put some kind of explicit "checkpoint" in coroutines or something like that?

that looks awesome! Bookmarked

For animations in my turn based game I took inspiration from that same "Coroutines in C" article, but went with an explicit context struct and manual checks. That code keeps getting called until exec->keep_calling is set to false (or rather, not set to true). It works, but is not a joy to write...

Right, saving the local variables is easy enough in lua, and can be done in C by using a struct, but the hard part is resuming at the correct place in the function. Extra-difficult if the code can change between save and load, which would disqualify the __LINE__ hack. But I like the macro tricks!

My biggest problem with them is thar it's hard to save the state of a program that has coroutines running (and resume them on load). That's the only thing preventing me from using them for e.g. cutscenes because I want to be able to snapshot the game at any point for debugging

yes, absolutely, people can do whatever they want. I gather you prefer not to have an "unset" / zero tile, and instead initialize an "empty" tilemap with the 1st tile of the palette? I can see cases where that would be useful. It's nice that there are no holes in the level then, too

I'm curious, what would you prefer? Zero for "empty" sounds good to me

You can download an iso straight from this page (scroll past the "Installation media")

Re-installing windows on my gaming PC (previous ssd died). Rufus is a godsend!

This was a good stream! - implemented the monkey tester in like 10 lines - let it run - it finds a bug (pointer to item in a growing array) - fix the bug, add a mode to detect similar bugs faster (force realloc for every element added to growing arrays)

This was a good stream! - implemented the monkey tester in like 10 lines - let it run - it finds a bug (pointer to item in a growing array) - fix the bug, add a mode to detect similar bugs faster (force realloc for every element added to growing arrays)

Oh, haha! Sorry for the rabbit hole, but glad my question was useful :) Makes a lot of sense to grow and eventually repack, I'm also doing that. I'm curious how you handle some specific aspects of resizing, I'll ask followup questions on your discord later!