Revert "ComputeFstat.h: FstatOptionalArgs doesn't take ownership of any pointers"
- Dec 02, 2021
-
-
Karl Wette authored
- This reverts commit 24114e27 - Change broke assumed behaviour of established code base, e.g. PyFstat
620971bd
-
This MR reverts one commit from !1712 (merged), which changed some memory semantics of the FstatOptionalArgs
struct in LALPulsar (i.e. whether the struct owns the memory of objects assigned to its fields). Reinhard has reported that this has broken existing code bases, e.g. PyFstat, which were assuming the previous memory semantics.
There's no "right" answer to what memory semantics the FstatOptionalArgs
struct should have; one could make arguments for different semantics depending on how one wishes to use the struct. That said, existing code shouldn't be broken without good reason (which there isn't here) and so this change is being reverted.
Restoring previous SWIG interface behaviour assuming by existing code base. Unlikely new code depends on the new semantics as this change was only recently pushed (and yet to be released).
n/a
- This reverts commit 24114e27 - Change broke assumed behaviour of established code base, e.g. PyFstat