tynet-lichat/shirakumo - IRC Chatlog
Search
20:02:53
shinmera
but if you like, once gjk+epa is done, there's plenty more algorithms to add (octtree, bvh3, grid2, k-d
20:17:36
shinmera
Hmmm, on that note, might actually be worth to add a serialisation interface to 3d-spaces.
4:34:55
|3b|
selwyn: any thoughts on using constants like xr:success and xr:error-handle-invalid instead of like :success and :error-handle-invalid ? (and maybe the same for other enums aside from result)
4:38:33
|3b|
ACTION is lazy so dislikes the extra 2 chars, and syntax highlighting could go either way, but particularly for the evolution model of khronos APIs :keywords are pretty annoying :/
4:40:27
|3b|
losing the type info is also somewhat annoying, but since that type info isn't hooked up into the editor and is frequently wrong at runtime anyway, i'm not sure how much weight to put on that
4:41:24
|3b|
ACTION considers doing the same for cl-opengl if i ever get around to making a modernized fork, since that has the same problems
6:45:58
|3b|
mostly a problem with return values, since user wants to compare them, and :whatever-foo isn't :whatever-khr isn't :whatever, but (= xr:whatever-foo xr:whatever-khr xr:whatever) could be true
6:47:27
|3b|
so we do extra effort to translate 5 to a keyword, possibly adding extra ERRORs if it happened to return a value we don't know about, only for the user to need to do extra work to figure out if that keyword mathes matches the one they are looking for
6:48:00
|3b|
where as if it just returns 5, user code compares that to a constant with the value 5, and it all works
6:50:16
|3b|
as far as type info part, thinking about it more that might actually be just a distraction and not a real concern, since we don't have any static typing (especially for cffi)
6:50:39
|3b|
so putting that type info into CL code just adds an extra runtime check that the foreign API is hopefully already doing
6:51:19
|3b|
and can be wrong (or outdated) on the CL side anyway, adding extra work if user wants to work around it without rebuilding bindings
6:52:12
|3b|
ACTION has the thought that code using CFFI could add on some static checks with a compiler macro or similar, but that's true with constants too
6:53:00
|3b|
if there is an automatic "turn negative return values into a CL ERROR" thing happening, that part can do the translation
6:53:10
selwyn
maybe the higher level bindings should be responsible for establishing restarts and so on