Smithay
Better data flow documentation
The data flow of Smithay is highly non-obvious. I was trying to understand how one would get a WlBuffer to render. It turns out that one obtains a WlBuffer from a BufferAssignment, which is in turn part of SurfaceAttributes. A SurfaceAttributes can be obtained from SurfaceData::data_map, which is part of the SurfaceData struct passed to the with_surface_tree_upward callbacks. Similarly, I had to figure out where the event loop happens (in Anvil, it turns out), and how to run it (create an instance of calloop::EventLoop manually). All of this took me about an hour to figure out, mostly by git grep -w on the source code.
This could definitely benefit from improved documentation. In addition, having a clear tutorial on how to make a really simple compositor would be nice. Anvil is over 5,000 lines of code, so it is a bit too big to be an example, especially for those (like me!) who mostly want a barebones compositor they can plug their own input and drawing backends into.
Yes that is fair. So far the code in our examples folder code was focused on the backend side of things. A simple compositor - likely only supporting the most barebones of protocols to render common apps - is a good idea.
This could definitely benefit from improved documentation. In addition, having a clear tutorial on how to make a really simple compositor would be nice.
Yes, absolutely, that is the longer-term plan of what the Smithay book ought to be. However it has not been in the focus lastly, as we were more focusing on actually getting everything to work. :sweat_smile:
