Use kDim instead of dim to avoid name collision#2802
Use kDim instead of dim to avoid name collision#2802umar456 merged 3 commits intoarrayfire:masterfrom
Conversation
…implementation in kernels.
umar456
left a comment
umar456
left a comment
There was a problem hiding this comment.
Thanks for debugging this and submitting the changes. Could you please rename the variable to kDim instead of _af_dim. I want to avoid making a variable name starting with an underscore which is usually reserved for system headers and compilers. Otherwise this looks good although this is will most likely break in the future because this is a bug that is only a problem in one platform. I think it would be better to to have some sort of assert in the header which will cause this to fail in any platform. I guess we can handle that in another PR.
glavaux2
changed the title
|
I have renamed to kDim which looks like working on Mesa AMD Clover |
| options << " -D To=" << dtype_traits<To>::getName() | ||
| << " -D Ti=" << dtype_traits<Ti>::getName() | ||
| << " -D dim=" << dim; | ||
| << " -D kDim=" << dim; |
There was a problem hiding this comment.
join.cl OpenCL kernel doesn't use this one at all. I wonder why it's being passed to this at all.
* Use kDim instead of dim to avoid name collision with some OpenCL implementations(AMD) in kernels. Co-authored-by: LAVAUX Guilhem <lavaux@iap.fr>
* Use kDim instead of dim to avoid name collision with some OpenCL implementations(AMD) in kernels. Co-authored-by: LAVAUX Guilhem <lavaux@iap.fr>
4 participants
This is related to #2800