darwin: add support for AppleM1 processor#225
Conversation
|
Please merge this to the master. I am facing the same problem. |
|
I also have this issue |
|
I tried adding a mac os ci stage: https://github.com/davisp/jiffy/blob/master/.github/workflows/ci.yml#L40 running on The CI passes currently passes on those workers, so wonder if this is still needed? |
I find hard to believe, but it could be that this got fixed in more recent llvm releases ? We already forked jiffy while back. I can't explain this result. Maybe the github docker images are multi-architecture but is running still on x86_64 🤷 |
🤷 I just went by https://github.com/orgs/community/discussions/102846 from Jan 13 this year:
I don't have an ARM MacOS laptop so can't test it locally. But something in llvm compiler being fixed sounds like a plausible explanation. |
TIL nice ! I tried it again notsureif... compare that to arweave (which uses the jiffy fork I mentioned with this fix) I don't know what to make of this, if this was the issue, been a while |
|
Thanks for checking @hlolli, much appreciated! Let's keep the PR open, it may need a longer investigation and may still be valid despite the CI on GH Actions passing. |
This was quite annoying thing to debug :) since port_compiler doesn't print link time errors, even with DEBUG=1
I tested by first cleaning
$ m -rf priv _build c_src/*.{o,d} c_src/double-conversion/*.{o,d}then in the rebar3 shell, I tested like seen in the screenshot
$ DEBUG=1 rebar3 shell