[R-SIG-Mac] (MacOS 15 x86_64) Error: running examples in ‘stats-Ex.R’ failed

[peter dalgaard]

If you goo further back, you will see FAIL on the same platform for quite a while, but failing in different places. E.g.,

2025-11-05T19:29:34.2882140Z   > coef(fmCc)
2025-11-05T19:29:34.2883410Z   (Intercept)           x         chB         chC         cha         chb         chc         chd 
2025-11-05T19:29:34.2885100Z    12.7781626  19.8494272  -0.8240301  -1.3309157 -31.7032317 -32.8819084 -33.3519985 -34.6249161 
2025-11-05T19:29:34.2886740Z   > ## (Intercept)           x         chB         chC         cha         chb         chc         chd
2025-11-05T19:29:34.2888830Z   > ##  12.7781626  19.8494272  -0.8240301  -1.3309157 -31.7032317 -32.8819084 -33.3519985 -34.6249161
2025-11-05T19:29:34.2891030Z   > (coef(fmCf) -> cf.f) # the same
2025-11-05T19:29:34.2892540Z   (Intercept)           x         chB         chC         cha         chb         chc         chd 
2025-11-05T19:29:34.2894330Z    12.7781626  19.8494272  -0.8240301  -1.3309157 -31.7032317 -32.8819084 -33.3519985 -34.6249161 
2025-11-05T19:29:34.2895830Z   > stopifnot(exprs = {
2025-11-05T19:29:34.2897020Z   +     identical(coef(fmCc), cf.f)
2025-11-05T19:29:34.2898910Z   + })
2025-11-05T19:29:34.2900180Z   Error: identical(coef(fmCc), cf.f) is not TRUE
2025-11-05T19:29:34.2902120Z   Execution halted

I think we have been here before and the culprit was stochasticity in the multithreaded Accelerate framework. Basically the examples do the same thing twice and get different results.

.pd

> On 17 Dec 2025, at 11.09, Jeroen Ooms <jeroenooms using gmail.com> wrote:
> 
> Checking base R on MacOS 15 fails a test on intel (not arm64), see:
> https://contributor.r-project.org/svn-dashboard/
> 
>   Error: running examples in ‘stats-Ex.R’ failed
>   ...
>   Error: identical(strip(qrI), strip(lmI[c("hat", "sigma")])) is not TRUE
> 
> _______________________________________________
> 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