|
Some functions use the same rational functions. For example, erfinv and erfcinv both use the same rational_p1q1, albeit through duplicate auto-generated folders. But besselj uses a totally different yet homonymous rational_p1q1. Shouldn't these homonymous functions be named differently so that factorization could be implemented in a robust fashion in the reference codebase? Or are there complex build-time mechanisms upstream that already take care of this? |
Replies: 1 comment
|
Using different names is unnecessary, as their usage is scoped to the relative module scope. Should these functions need to be present in the same namespace, similar to a C compiler, one can rename symbols when inserting into a common namespace. We do this already when generating "flat" bundles in which all functions exist in the same top-level scope. |
Using different names is unnecessary, as their usage is scoped to the relative module scope. Should these functions need to be present in the same namespace, similar to a C compiler, one can rename symbols when inserting into a common namespace. We do this already when generating "flat" bundles in which all functions exist in the same top-level scope.