[R-SIG-Mac] Which version of clang does CRAN's R build use?

[peter dalgaard]

For whatever it is worth, the nightly source tarballs are built with the latest xcode CLT (because I fell into Apple's 16.3 trap and didn't bother with downgrading to 16.2). 

Of course the resulting binaries don't get published anywhere, but they seem to pass all tests.

-pd

> On 1 Sep 2025, at 01:58 , Simon Urbanek <simon.urbanek using R-project.org> wrote:
> 
> Duncan,
> 
> 
>> On 1/09/2025, at 08:25, Duncan Murdoch <murdoch.duncan using gmail.com> wrote:
>> 
>> On 2025-08-31 4:11 p.m., Duncan Murdoch wrote:
>>> The instructions on https://mac.r-project.org/tools/ say to install
>>> tools using `sudo xcode-select --install`.  Currently for me that installed
>>>   Apple clang version 17.0.0 (clang-1700.0.13.5)
>>> Is that the version of clang that is being used in the CRAN builds?  If
>>> not, is there an advantage to matching what was used?
>> 
>> Just found on the download page for R 4.5.1 "This release uses Xcode 16.2(arm64)/14.2(x86_64) and GNU Fortran 14.2."  So that answers my first question, but leaves the second one.  Is it worth installing that version?
>> 
> 
> It depends - if you want stability then, yes, since there were major breaking changes in Xcode 16.3 (Apple clang 1700). Xcode 16.2 was the best "stable" release before that. Similarly, if you want produce binaries for others then you want it for the same reason. However, if you want to test the cutting edge and test support for new standards then using the latest is ok. The fallout isn’t huge, it depends on what you will try to build.
> 
> Cheers,
> Simon
> 
> _______________________________________________
> R-SIG-Mac mailing list
> R-SIG-Mac using r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-mac

-- 
Peter Dalgaard, Professor,
Center for Statistics, Copenhagen Business School
Solbjerg Plads 3, 2000 Frederiksberg, Denmark
Phone: (+45)38153501
Office: A 4.23
Email: pd.mes using cbs.dk  Priv: PDalgd using gmail.com