Issues should be filed on github.
You can also follow the project on cohost
My workflow is in fossil, so the github repo might be a bit outdated, eventually i should implement some sort of automatism to keep the repos in sync, but for now i am ok with doing it by hand every week or so. Whenever i make some progress i will make a commit on github :D
For "bleeding egde", clone this fossil repo as below.
fossil open https://www.awesemble.de/cgi-bin/fossil/public/wjpu --workdir wjpu --repodir ..
will give you a similar experience to
git clone https://github.com/kamisori/wjpu
except you will end up with a wjpu.fossil file in the folder above the workdir.
i like to put a
.fossil folder containing all the files like:
*.fossil somewhere like the root of the folder holding the workfolders. This way the workfolder isn't cluttered with database files. there is actually no requirement for the
*.fossil fileextension at this time, so you could name it .fossil.reponame instead and have it hidden and also "namespaced" away in autocomplete but in the same folder still.
i know it doesnt matter if you give a program a path to a directory with or without the ending slash/backslash, but it just ticks me off when autocomplete bumps into the databasefile's dot in its name when cd-ing from my work dir into a projects workdir, or at all on autocomplete in the shell... i dont want to interact with the fossil databasefile directly (most of the time).
for the purpose of this document i will refer to this projects root as
Download the wgpu distribution for 'any' platform to use this projects jpm file. see also eliemichels tutorial on webgpu which i used to write all this.
copy to your ./wjpu/ and extract like e.g. this:
you should end up with a folderstructure like:
also the webgpu-release compatibility shim to make a transition to dawn easier in the future. Put this file into:
git clone the following into ./wjpu/:
in ./wjpu/ execute:
cmake -S ./glfw/ -B ./glfw/build/ cmake --build ./glfw/build/
jpm install spork
clone repo and position yourself into its root, then run the c code generating rule before building to get fresh bindings made out of cjanet code:
jpm run gen jpm build jpm test
on windows you could just call jake.bat and it'll do all that in one go. i added a jake.sh for similar convenience on linux