Parsing month names fails when day not present.

[coolbutuseless]

parse_date_time("2 july 2008", orders="dmy"); parse_date_time("7 2008", orders="my"); parse_date_time("july 2008", orders="my")
[1] "2008-07-02 UTC"
[1] "2008-07-01 UTC"
[1] NA
Warning message:
All formats failed to parse. No formats found.

[coolbutuseless]

also I would have thought these would return the same date:

dmy("1 7 97"); dmy("1 july 97")
[1] "2097-07-01 UTC"
[1] "1997-07-01 UTC"

[coolbutuseless]

And now I've just discovered that strptime() also won't properly parse month names when day not present:

strptime("JUL 97", "%b %y")
[1] NA
strptime("1 JUL 97", "%d %b %y")
[1] "1997-07-01 EST"

[vspinu]

For non-numeric formats (months in this case) lubridate relies on strptime. For numeric formats an internal C parser is used. So you get NA on "july 2008" and the correct answer for "7 2008".

Same reason for dmy("1 7 97") and dmy("1 july 97") discrepancy. It looks like strptime parses all years greater than 68 as XXth century. I will look into fixing this asap.

I thought about extending the internal C parser to handle alphabetic months in English locale. It might be a worthwhile effort given this common user pattern.

[vspinu]

Fixed the 68 issue and openning a new ticket for C parser of %b format.