every() and NAs

[kablag]

I'm very impressed by purrr but from my point of view it's strange result:
every(NA, function(x) x > 0)
[1] FALSE
Doesn't it break R rules about NA?
Maybe it's better to return NA like all()?
all(c(NA) > 0)
[1] NA

[lionel-]

hmm yes, we should probably return NA if one element of .x is NA. And with some(), we should do the same except if one element is TRUE.

[hadley]

library(purrr)
every(NA, ~ .x > 0)
#> [1] FALSE