[R-SIG-Mac] macos 26 considerations

[Peter Dalgaard]

Hi Simon,

As you know, the Apple antics resulted in the release-build machine getting ahead (clang 17) of the machines that builds the CRAN prereleases.

This mostly annoying for me since I only distribute source tarballs, so as long as it doesn't actually break compilation, we're fine. However, I can't use the CRAN package binaries, so everything needs to be compiled from source. And if I want to be able to (say) R CMD check on packages that act up on CRAN, sometimes they install everything including the kitchen sink via dependencies. 

I had planned to update my setup in the interim between 4.5.3 an 4.6.0. It might be a good idea to try and synchronize this time.

I take it that going to Tahoe + Xcode CLT 26.3 should be OK?

- Peter

> On 4 Mar 2026, at 02.38, Simon Urbanek <Simon.Urbanek using r-project.org> wrote:
> 
> Bill,
> 
> on that front, Apple has again been making large breaking changes in Xcode (which they are often slow to fix; with Xcode 26.4 they are going from clang 19 to 21) so I’d definitely avoid going past Xcode 26.3 as that is too risky. That said, I think most of the growing pains in Tahoe have settled (you can pretty much consider x.0 alpha and x.1 beta with most macOS major releases).
> 
> Cheers,
> Simon
> 
> 
>> On 4 Mar 2026, at 13:27, Bill Evans via R-SIG-Mac <r-sig-mac using r-project.org> wrote:
>> 
>> Gabor,
>> 
>> Even with macos 26.2, are you on xcode-14.3 and gfortran-14.2? Or are you working in R-devel and therefore on xcode-14.4?
>> 
>> Bill
>> 
>> On 2026-02-22 11:47, Gábor Csárdi wrote:
>> 
>>> On Sun, Feb 22, 2026 at 6:25 PM Bill Evans via R-SIG-Mac
>>> <r-sig-mac using r-project.org> wrote:
>>> [...]
>>>> On 2026-02-22 09:15, Gábor Csárdi wrote:
>>>> 
>>>>> I am 26.2 and everything works fine, no need to reinstall anything,
>>>>> except maybe xquartz, which sometimes needs a reinstall after an
>>>>> update. But xquartz works fine as well.
>>>> Very interesting! I’m curious about the difference in your experience
>>>> and Prof Murdoch’s, perhaps the rgl-dev work is more nuanced than you’ve
>>>> tested in your use of xquartz? I wonder if xorg-server would be any
>>>> better (since it is quite a bit newer) for that.
>>> I don't use rgl or x11, so I indeed only ran `x11(); plot(1:10)` for a
>>> minimal test, and opengl indeed does not seem to work if I try
>>> library(rgl).
>>> 
>>> If I just install xorg-server and xinit, then R does not find the X
>>> server when loading the R_X11 dll.
>>> 
>>> G.
>> ​
>> [[alternative HTML version deleted]]
>> 
>> _______________________________________________
>> R-SIG-Mac mailing list
>> R-SIG-Mac using r-project.org
>> https://stat.ethz.ch/mailman/listinfo/r-sig-mac
>> 
> 
> _______________________________________________
> 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