split project into static and dynamic modules
This commit is contained in:
parent
55dca44b4a
commit
b6b9054d0d
28 changed files with 188 additions and 161 deletions
15
README.md
15
README.md
|
@ -35,6 +35,7 @@ you will have to build. The specific version we built against is https://github
|
|||
|
||||
* MacOS 10.12.6 - Python 3.7
|
||||
* Ubuntu 18.04 LTS - Python 3.6
|
||||
* Windows 10 (64 bit) - Python 3.7
|
||||
|
||||
# HELP WANTED
|
||||
|
||||
|
@ -43,10 +44,10 @@ you will have to build. The specific version we built against is https://github
|
|||
|
||||
### Dynamic bindings
|
||||
|
||||
I have attempted to do CFFI ABI dynamic bindings too in order to avoid the need to compile a C extension module,
|
||||
but they don't work properly. They fail in the same place the ctypes binding fails, accessing
|
||||
materials of a model, because Python can't dynamically tell the difference between a pointer and an array. There's probably
|
||||
some way to specify this (e.g. in raylib_modified.h) but it's difficult to be sure we fixed them all because the errors
|
||||
are often completely silent.
|
||||
|
||||
See test_dynamic.py for the non-working example.
|
||||
In addition to the API static bindings I have attempted to do CFFI ABI dynamic bindings in order to avoid the need to compile a C extension module.
|
||||
There have been some weird failures with dynamic bindings and ctypes bindings before and often the failures are silent
|
||||
so you dont even know. Also the static bindings should be faster. Therefore I recommend the static ones...
|
||||
|
||||
BUT the dynamic bindings have the big advantage that you don't need to compile anything to install.
|
||||
|
||||
See test_dynamic.py for how to use them.
|
Reference in a new issue