Fail instead of trying to cast Inf to int

[Enchufa2]

Closes #1455.

min(integer(0)) shows a warning and results in a numeric value. We cannot do that (because we keep the same type in storage instead of returning a SEXP), so I think that it's best to just throw an error in such cases.

Checklist

• Code compiles correctly
• R CMD check still passes all tests
• Preferably, new tests were added which fail without the change
• Document the changes by file in ChangeLog

[eddelbuettel]

This looks good. Do you think you can toss in a check for REAL returning limits as expected?

[eddelbuettel]

Sorry about 'da noise'. Looks good.

[Enchufa2]

Thanks, merging this now, will rebase the other one after revdep checks.

[eddelbuettel]

Well these are 'behavior changes' (that are obviously merited) so we need rev.deps (in case we need to talk affected maintainers). "Moar cpus" would be good; back on planet earth we just wait.

[Enchufa2]

Ops, you are right, I should have waited. In my defense, if someone was using min/max on an empty integer vector, this would have been flagged on CRAN by UBSAN checks, so it is extremely unlikely that there are any issues in revdep checks after this change.

[eddelbuettel]

Agreed. And if there is something we will catch it in the next wave of tests.