Re: strange behaviour with strtotime()

Ford, Mike wrote:

Uhm -- my mistake, sorry -- 31st Feb equates to 3rd March, not 2nd, so 3rd April here is correct. The basic thing is that strtotime() just uses the same day-of-the-month when calculating in months -- so, on the 31st January, it reckons thus: +1 month: 31st February -- *doesn't exist* converted to 3rd March

+2 months: 31st March    -- OK
+3 months: 31st April    -- *doesn't exist* converted to 1st May
+4 months: 31st May      -- OK
+5 months: 31st June     -- *doesn't exist* converted to 1st July

etc., etc. So you just have to be careful when calculating next/previous months to allow for strangeness like this near the end of a month -- just as you have to watch out for DST transitions if calculating with times near 01:00/02:00/03:00. If all you're interested in is the actual month, your best bet is to normalize to a date that exists in every month (such as 1st) before doing your next/previous calculation. Your table is quite ok.

But converting dates just because it doesnt exist and on the next added month doing so as if there was everything ok is a bug for me. 31st January -> +1 month -> 31st February (not possibele so its 3rd February) -> +1 month -> 31st February (What the O_o) How does he calculate? For me 3rd Fabruary -> +1 Month -> is 3rd March also 31st January -> +2 Month -> is for me 3rd March 31st February is not true because i miss "3" (!!) days. Not for you? Barry -- Smileys rule (cX.x)C --o(^_^o)