CRAN Package Check error on Solaris only

[jmgirard]

I am getting a CRAN Package Check error for a test on Solaris only (not on Windows, MacOS, or other Linux platforms). It seems like that discrepancy is small enough to maybe be due to differences in rounding. But this is also an estimate from a bootstrap distribution so it could perhaps be due to differences in random number generation (I am using set.seed() earlier in the test). It's hard to troubleshoot this issue because I don't know anything about Solaris. Any hints or tips would be greatly appreciated.

round(res$results$e_uci, 3) (actual) not equal to 0.708 (expected).
actual: 0.709
expected: 0.708

https://cloud.r-project.org/web/checks/check_results_circumplex.html

[technocrat]

Solaris subsists to this day by grace of Oracle, I suppose, as a sort of Galapagos Island version of Unix. It is said to generate far more fails than Linices due to a particularly severe view of C++. See this review, which may still be apropos.

[maelle]

You might find this blog post by @Gabor, " Checking your R package on Solaris", relevant.