Martin Kinkelin

I guess `compilable/fix19983.c` needs a working #4491. **Edit**: And #5032.

The original report is about Termux's `cc` as linker driver (for `ld.bfd`, which isn't supported anymore for recent LDC versions - needs the lld linker, like recent NDKs) - we...

Thx! I don't really recall whether the GC2Stack regressions were the reason, but they are definitely a problem. - We have an extra LLVM pass that looks for GC allocations...

Once everything GC-related is templated (or at least the stuff that makes sense), the GC2Stack would need to be overhauled, incl. accounting for the new unstable mangled names and populating...

I guess what's missing for LLVM to auto-vectorize these ops is the `@restrict` (well, at least in such an isolated case, without it being inlined and the `restrict` potentially being...

https://dlang.org/spec/arrays.html#array-operations, 13.12.6: > The slice on the left and any slices on the right must not overlap. All operands are evaluated exactly once, even if the array slice has zero...

Thx for going down yet another rabbit hole! :) - The posix fixes will need to be upstreamed. Weird that one CircleCI job fails: ``` Testing memoryerror_null_read /home/circleci/build/runtime/druntime-test-exceptions-debug/memoryerror_null_read 2>/home/circleci/build/runtime/druntime-test-exceptions-debug/memoryerror_null_read.stderr ||...

That just enabled the doubledouble type, if the C++ compiler uses it on that platform. So the issue remains in upstream Phobos, probably just a simple constant missing for that...

Oh sorry, just checked your diff (manually) - so the propagation of the host-C++ `long double` is only enabled on non-musl Linux, not for any BSDs etc. Which is probably...

The installer only provides the arm64 libs for cross-compilation; the binaries are x64 only.