Fixed to allow compilation on macOS M1 #49
Conversation
|
We use this pull request in conda-forge (see #93) - is there a reason why it is not being merged? |
|
I think the changes around FindPythonPyEnv are not needed, but the other ones are relatively trivial and are required for clang. |
|
Thanks @ducha-aiki for the PR. We'll merge only if the changes are minimally required for the feature - it looks like right now there they're fixing multiples issues at once. Are these changes related to M1 architectures or to clang? I don't have an M1 computer at hand but I can run clang. |
|
Well, for me, the "feature" is "works on my M1 machine with minimal changes". I don't have any Intel clang, unfortunately. Let me check if that works w/o FindPythonPyEnv and with auto-AVX |
|
@skydes I have updated PR |
|
LGTM :) |
| // The following two Evaluate overloads are needed for interfacing | ||
| // with automatic differentiation. The first is for when a scalar | ||
| // evaluation is done, and the second one is for when Jets are used. | ||
| template <> |
There was a problem hiding this comment.
Explicit specialization in class scope is accepted by clang but not by GCC:
error: explicit specialization in non-namespace scope ‘class pixsfm::Interpolator’
This is part of the standard only since C++17 and still a bug in GCC: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=85282 Will look into this when I find the time.
Full story described here #47