-
Notifications
You must be signed in to change notification settings - Fork 417
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Mac CI using single-precision libccd #291
Comments
I think you are correct, though the homebrew formula for libccd is trying to use double precision: I'm investigating |
You have no idea how overjoyed that statement makes me. |
It looks like the libccd formula is using the latest tag In the short term, we should update the libccd formula with the corrected cmake argument. We can also ping the libccd maintainer to see if they can make another release. cc @jslee02 |
Being largely mac ignorant, I'm going to cast my vote modulo what actually works. I like the flavor of not having to wait for libccd but I also like the idea of keeping the simplest of dependent relationships. So, I'd vote:
|
This needs a refresh. The homebrew formula now specifically refers to libccd v2.1. However, between libccd v2.0 to v2.1, the CMake mechanism for specifying double precision changed from the macro So, it seems we've reversed our old problem where, originally, we were pulling v2.0 but using the v2.1 spelling. Now we're using v 2.1 but using the v2.0 spelling. I'm not a mac user, so I'd ask for one of the previous supporters of mac to help out on this one. If no one volunteers to do so, I'll step in. @scpeters @jslee02? |
PR is submitted: Homebrew/homebrew-core#37365 |
It doesn't make a difference in the 2.1 release. Both ways of specifying precision are supported: |
Given that with that observation and all of the discussion in the homebrew pr, we can safely close this issue. (Feel free to re-open this if you feel I've been overly optimistic.) |
It appears that the mac ci is using single-precision libccd.
#290 failed the new unit tests. Instead of getting solutions to a tolerance of 1e-14, it was getting results satisfied to within 1e-7. Furthermore, when evaluating the tests on a local mac with known double-precision libccd, these errors did not appear. But when forcing it to use single-precision libccd, the errors appeared.
The text was updated successfully, but these errors were encountered: