In your vignette and tests, you use rpois(100, 5e6) along with rnorm(5e6). The call to rpois is producing only 100 random samples; data.frame won't complain because 100 cleanly recycles into 5e6.
(I doubt this has any influence on the benchmark or results, but the code is inconsistent.)
In your vignette and tests, you use
rpois(100, 5e6)
along withrnorm(5e6)
. The call torpois
is producing only 100 random samples;data.frame
won't complain because 100 cleanly recycles into 5e6.(I doubt this has any influence on the benchmark or results, but the code is inconsistent.)