aboutsummaryrefslogtreecommitdiff
path: root/platform (unfollow)
Commit message (Collapse)AuthorFilesLines
2023-07-26macOS fix memory leak (hopefully)HEADmasterNicholas Tay1-1/+1
I think it's this simple, maybe I got confused by the memory going up a bit. But I think this actually works, simply releasing should ARC cleanup, and the memory goes up a bit just because of the NSCache but doens't go up further. Again it'd be nicer to use some different lower level way of playing, but this is definitely good enough for now.
2022-12-09Initial working Obj-C sound (leak)Nicholas Tay3-12/+71
It seems to memory leak so needs to be fixed, but this was before I had to return my work MacBook - until I get my own!
2022-11-03Initial macOS support (key hooking only, no sound yet)Nicholas Tay1-0/+167
Thought I'd commit this first, since it'll be some Objective-C stuff coming (tested it out in another mini probe). Mostly has stuff similar to Linux so probably should abstract some out to *nix common. But uses IOKit to get into the typing events. Will use Obj-C for now for sound, since it'll probably be a NSSound thing, hooked up with NSCache (like Linux) to reduce loading into NS format multiple times. Also probably would need to free the sound object on finish sound, and not sure how I'd do Obj-C delegates from C... it would be fun to figure it out though eventually Also switched the Makefile to use clang, it's warnings do seem to be nicer :) and is what `gcc` is aliased to on a Mac by default anyway.
2022-05-30Add a proper cache for Linux+SDLNicholas Tay1-14/+50
2022-05-18Fix indent/stylingNicholas Tay2-12/+24
2022-05-17Add Linux supportNicholas Tay3-4/+169
Oh boy this was a bit of a hassle lol - dynamic loading was the easiest part... but then came both sound and x11 Using SDL for now but I'd really like to change it for even lower layer, but then I might have to make my own mixer... oh no.
2022-05-11Improve plugin loading architectureNicholas Tay2-16/+11
This should be better, maybe the variable names could be better though. Init is more extensible, we take in a 'board_data' struct on the plugin end that has everything it needs (mainly function pointers for now). Then, a 'board' struct is given back to the main Clak runtime, with everything it needs to know (again, mainly function pointers). It is a bit weird that the board is not stored with Clak but as a pointer to the dynamically loaded bit, but not sure.
2022-05-10Dynamic load boards as DLL (windows only for now)Nicholas Tay2-0/+43
The .h files are pretty weird, should look at other C projects to see how they load plugins. This function pointer business with typedefs is kinda weird, not sure where they should live.
2022-05-07Split out platform-specific code for WindowsNicholas Tay2-0/+90
Not sure if I'm really doing it in the best way possible. Feels a bit weird that some place assumes the existence of other functions.