Re: sdl2 insert/delete and F14

[Mete Balci]

Yes I saw, the API is stable but it is a preview release.

I will add SDL3 support and want to work on it a bit and see the other option about kbd mapping (removing X11 dependence from SDL backends).

On Tue, Nov 26, 2024 at 8:33 AM Björn Victor <bjorn@victor.se> wrote:

On 2024-11-26 08:26, Mete Balci wrote:

I can add sdl3 support but I think it is still considered as preview not stable.

From https://wiki.libsdl.org:

SDL3 has officially stabilized the API and ABI, and a preview release is available at the GitHub page. We will still be adding things to the API but not changing or removing anything, so it's safe to start porting your projects from SDL2.

so it should be OK. If you look at the SDL2 documentation, it says (in the "leafs"):

(This is the legacy documentation for SDL2, the previous stable version; SDL3 is the current stable version.)

But I haven't checked out the differences.

One issue at the moment is even if X11 backend is not used, because of kbd mapping, X11 headers/functions are required. They do not involve any display operation but still I was not expecting that when I disable the X11 backend. To eliminate this, I think the only solution is to "invent" a backend independent kbd mapping. Then, if X11 backend is used, kbd mapping will be from Xkeys to LM keys, and if SDL backend is used, it will be from SDL scancodes (or whatever is appropriate) to LM keys. This is cleaner in one sense, but complicated in another sense. An obvious benefit is when SDL is used, X will not be involved anywhere (no key sym/code transformation from SDL event to X).

I think it would be ambitious and difficult to have a backend independent mapping - there is also a risk that you can only use the "least common denominator", i.e. only things that are supported by all backends, even if one has better support?

I totally agree on the benefit of removing X11 dependence!

-- Björn