Closed danielinteractive closed 8 months ago
Yes, that would be super helpful for my package, too (and thanks also from my side for rhub2, which has helped me a lot with package testing).
This could be helpful: @gowerc spent quite some time this week to spin up this Docker file producing a clang18 system. This resembles very closely (most likely not identical of course because we are missing details from CRAN) the CRAN system's setup.
Thanks, that Dockerfile
might very well be needed in the future, but for now I want with a simpler way using binary compiler packages provided by the LLVM project. They seem to reproduce the issues CRAN found.
Building the container right now: https://github.com/r-hub/containers/actions/runs/7904324193
Once it is deployed I'll add it to the rhub2 packages as well.
All done. Please open an issue here or in the r-hub/containers
repo if you have any issues.
Thanks a lot @gaborcsardi !
Hey @gaborcsardi, Thank you so much for working on this! For personal interest how did you get binaries for clang18 for Fedora? I tried searching but the only ones I could find were those provided by their package manager but they appeared to be locked to clang17.
It is not fedora, it is ubuntu: https://r-hub.github.io/containers/containers.html#clang18 https://github.com/r-hub/containers/blob/main/containers/clang18/Dockerfile
It doesn't matter much what distro it is.
Ah ok, interesting. I would say for our mmrm package we've run into few runtime issues with our Cpp code that only came up on CRANs fedora-devel machine. Thats mainly what motivated us to try and really replicate that machine as closely as we could. Though to your point I never really understood why the issues only came up just that machine as I would have expected the compiler version to be the problem rather than the OS (perhaps it was just an unfortunate coincidence).
Yes, I would be very surprised if the difference is not due to different compiler options, or maybe versions. Which version of mmrm was this?
I'm struggling to remember but I think it was v0.2.2 when we first started getting these fedora-devel-clang issues, @danielinteractive will hopefully be able to confirm.
Yeah that should have been about that time / version.
Hi all,
thanks a lot for
rhub2
, looks great!One feature request that you might already be working on: could we get a
clang18
system that is resembling the additional checks on CRAN (I believe by Prof Ripley) closely? That would be super helpful.Thanks, best regards Daniel