You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
That class is an atrocious abomination and belongs on a trash heap!
It is annoying to use, even more cumbersome to actually do something with it and it poisons the cleaness of other classes like PointerPointer and similar.
So it should be done better before an official release. 3.0.0 can be realesed, but the new jnigen binder/runtime should be marked as incubating.
The idea of that class is to bypass the different sizes of c int types that can arise with different compilers/architectures.
It currently does so by infering over the compiler the actual size of the type it is supposed to wrap.
I think a better approach is to have pointer types that are
either variable size types like IntPointer and we make assumption of the size of an it per architecture. When compiling, we can statically assert that our assumption is correct.
Have fixed size pointer like UInt64Pointer.
This should hopefully make the pointer easier to work with while retaining guarenteed correctness.
Furthermore, I'm tempted to make all Pointer types extend from VoidPointer.
Also, EnumPointer always assume an enum size of "int", which is incorrect.
The text was updated successfully, but these errors were encountered:
That class is an atrocious abomination and belongs on a trash heap!
It is annoying to use, even more cumbersome to actually do something with it and it poisons the cleaness of other classes like PointerPointer and similar.
So it should be done better before an official release. 3.0.0 can be realesed, but the new jnigen binder/runtime should be marked as incubating.
The idea of that class is to bypass the different sizes of c int types that can arise with different compilers/architectures.
It currently does so by infering over the compiler the actual size of the type it is supposed to wrap.
I think a better approach is to have pointer types that are
This should hopefully make the pointer easier to work with while retaining guarenteed correctness.
Furthermore, I'm tempted to make all Pointer types extend from VoidPointer.
Also, EnumPointer always assume an enum size of "int", which is incorrect.
The text was updated successfully, but these errors were encountered: