cumany and missing values

[hadley]

cumany()/cumall() should always be NA after the first NA is seen. Currently:

cumany(c(NA, NA, NA))
# [1]   NA TRUE TRUE
cumall(c(NA, NA, NA))
# [1] NA TRUE TRUE

[romainfrancois]

Hmm. What should happen if the NA appears where we already know the final answer:

cumall( c(FALSE, FALSE, NA, TRUE)  )
cumany( c(TRUE, TRUE, NA, FALSE ) )

Those should be all FALSE and all TRUE IMO. So in cumany as soon as we see TRUE or NA we know everything, and same in cumall as soon as we see NA or FALSE.

[hadley]

Right - they should follow what R does - TRUE || NA is TRUE.

[romainfrancois]

Cool. Closing this then :)