Unnecessary warning from gmean about sum greater than integer can hold

[mattdowle]

See at 3:01 into this video :
https://www.youtube.com/watch?v=AmE4LXPQErM

[jangorecki]

data.table(a=c(1L,1L), v=c(2e9L, 2e9L))[, mean(v), a]
#       a    V1
#   <int> <num>
#1:     1 2e+09
#Warning message:
#In gmean(v) :
#  The sum of an integer column for a group was more than type 'integer' can hold so the result has #been coerced to 'numeric' automatically for convenience.

@mattdowle is it fine to add extra argument to gsum called silent or ignore_overflow?

[HughParsonage]

Or just suppress the warning altogether? The average of an integer vector isn't an integer anyway, so avoiding coercion is not a concern.

[MichaelChirico]

@jangorecki yea it seems like this is the most expedient way in the current API. Problem is that is a registered routine from R so it needs a bit more effort.

What would be easier is if we have access to a suppressWarnings function from the C level, and we could just wrap gmean's gsum call here:

ans = PROTECT(suppressWarnings(gsum(x,narm))); protecti++;

I don't think that's possible though? So a new argument is the way to go...