Parse format "%Y%m%d" (no delimiter between fields) about cctz HOT 3 CLOSED

Right ... %Y will happily continue to consume digits for the year, and there is no backtracking to retry failed matches with less greediness.

If you know that your year spans exactly four digits, use the %E4Y specifier. That is "%E4Y%m%d" will successfully parse "20140304" as 2014-03-04.

Aside: "%Y%m%d" on "20140304" doesn't return 1970-01-01 ... it fails (returns false) and leaves the output time_point alone.

from cctz.

Great to know, thank you!

I wonder whether this may possibly become a somewhat frequent issue when adopted as standard ("%Y" is quite common after all, as are time formats without delimiters), the format strings might better be kept backwards-compatible to strftime()/strptime(), and extensions should be pure extensions, without breaking traditional, expected behavior?

A possible backwards-compatible format could be:

"%Y" -- strptime()-like four-digit year YYYY
"%E#Y" -- Year with # digits of fractional precision
"%EY" -- Year with arbitrary fractional precision

What do you think?

from cctz.

Hi,

Thanks for the suggestion. We agree with you that this might be a not-totally-uncommon transition issue. However, we also think that the current greedy %Y is the "best" behavior.

So, for the time being we'll add something more prominent to time_zone.h where %Y is described. However we will certainly raise this question during any standardization process.

Thanks again.

from cctz.