shelvick
commented
1 year ago Hi there, thanks for your enthusiasm and willingness to push the project forward! I'll say that overall the changes you're proposing are more work than I (personally) am willing or able to do right now -- but that doesn't mean I wouldn't merge any PRs you submit, as long as they make sense. You could also of course make a fork; no offense taken.
It may be helpful though to understand some of the design choices before you go on that path:
Raylib and Raygui are merged into Claylib as a single wrapper library.
Yes, the vision of Claylib was as a beginner-friendly "batteries included" tool, which meant including things like Raymath and Raygui and wrapping them in a strong Lispy layer. cl-raylib exists for people who might want just a thinner wrapper over Raylib only. I suspect it wouldn't be too difficult to split into something like Claylib and Claygui. Where you'll run into trouble is with Raygui itself, because you actually have to build Raygui differently depending on whether you're using it with Raylib or standalone. That would in turn cause issues using Claylib and Claygui at the same time.
The Raygui executable is bundled with the wrapper instead of being installed separately. This is slightly inconvenient for me as I would prefer to install the executable myself.
Claylib does support this, via the CLAYLIB_USE_SYSTEM_RAYLIB_LIBRARIES environment variable. See #71. Though, do also see #77 -- it expects the tagged Raylib 4.5 release.
Probably due to the second point, Claylib is not available in Quicklisp. I have to clone the repository into local-projects.
That's not due to the second point, but the fact we've only tested comprehensively on Linux and SBCL. This is relatively low-hanging fruit and I'd love to be able to build for Mac and Windows, and ideally other CL distributions too.
I am not satisfied with the implementation of vectors and matrices. In the Lisp community, there are very successful libraries for 3d-vectors, 3d-matrices, and 3d-quaternions (let's call them the "linear trio"). The function names in these libraries are much more convenient (although the names are shorter, so it will take some getting used to). It would be great to use them instead of reinventing the wheel.
We didn't reinvent the wheel -- not exactly. What we did was create "pass-through" CLOS objects. If you look under the hood you'll see that they're really just convenient ways of interfacing with Raylib's own C structs. I'm not entirely familar with how those Lisp libraries work so am willing to be proven wrong, but I suspect it won't be as easy as you're hoping to just plug those in and have them work seamlessly with the passthrough Raylib functions. For Claylib's own native shape classes, maybe, but then that seems like more trouble than it's worth.
Filipp-Druan
Hello! I like a Raylib but Claylib not everything suits me: 1) Raylib and Raygui are merged into Claylib as a single wrapper library. 2) The Raygui executable is bundled with the wrapper instead of being installed separately. This is slightly inconvenient for me as I would prefer to install the executable myself. 3) Probably due to the second point, Claylib is not available in Quicklisp. I have to clone the repository into local-projects. 4) I am not satisfied with the implementation of vectors and matrices. In the Lisp community, there are very successful libraries for 3d-vectors, 3d-matrices, and 3d-quaternions (let's call them the "linear trio"). The function names in these libraries are much more convenient (although the names are shorter, so it will take some getting used to). It would be great to use them instead of reinventing the wheel.
I would like to suggest splitting the library into two separate libraries. Additionally, I would like to replace your implementation of linear algebra with the "linear trio" libraries that are widely used in the Lisp community. Do you think it would be difficult to make these changes? I'm ready to write new libraries on my own.
Thanks!