[flang-commits] [flang] [flang] Decouple HLFIR array expansion from optimization level (PR #233478)

[via flang-commits]

================
@@ -103,13 +103,12 @@ def InlineHLFIRAssign : Pass<"inline-hlfir-assign"> {
                      /*default=*/"false",
                      "Only inline assignments where RHS is a numeric or "
                      "logical scalar broadcast "
-                     "to an array LHS. Used at O0 to avoid runtime calls in "
-                     "device code without inlining all array assignments.">,
+                     "to an array LHS, even when other array operations "
+                     "are left to the runtime.">,
                  Option<"onlyCUDADeviceContext", "only-cuda-device-context",
                         "bool", /*default=*/"false",
                         "Only inline assignments in CUDA Fortran device "
-                        "code. Used at O0, where a module holds both host "
-                        "and device subprograms, to leave host code alone.">];
+                        "code, leaving host code unchanged.">];
----------------
jeanPerier wrote:

Removed the option, helper, both guards, and the obsolete comment.

https://github.com/llvm/llvm-project/pull/233478