discrepancy between strptime and ymd

[markvanderloo]

ymd's doc states that it is based on 'strptime'. However, for the edge case of a non-existing date I get different results.

> ymd("2011-02-29")
[1] "2011-03-01 UTC"
> strptime("2011-02-29","%Y-%m-%d")
[1] NA

[vspinu]

ymd's doc states that it is based on 'strptime'.

This used to be true but it's not anymore. ymd relies only strptime when non-numeric formats are used. Otherwise a fast internal parser is used. This is documented in parse_date_time:

 ‘parse_date_time’, and hence all the derived functions, such as
 ‘ymd_hms’, ‘ymd’ etc, will drop into ‘fast_strptime’ instead of
 ‘strptime’ whenever the trained from input data formats are all
 numeric.

I have just fixed this in the docs. Thanks.

ymd("2011-02-29") ## [1] "2011-03-01 UTC"

This is a semi-bug. I will fix this in the parser if it doesn't impose too much of the penalty. You are expected not to pass invalid dates to the parser.