问题
Here's the simple example (today is 2016-08-29):
var_dump(Carbon::now());
var_dump(Carbon::now()->addMonths(6));
Output:
object(Carbon\Carbon)#303 (3) {
["date"] => string(26) "2016-08-29 15:37:11.000000"
}
object(Carbon\Carbon)#303 (3) {
["date"] => string(26) "2017-03-01 15:37:11.000000"
}
For Carbon::now()->addMonths(6) I'm expecting 2017-02-29, not 2017-03-01.
Am I missing something about date modifications?
回答1:
There is no 2/29/2017. Leap-year happened in 2016.
The following link:
http://carbon.nesbot.com/docs/#api-addsub
provides an example of adding 1 month to 1/31/2012 and arriving on 3/3/2012. Which is intended, though seems confusing to me.
As a counter-example exhibiting different behavior, in SQL:
SELECT dateadd(m,1,'2012-01-31')
results in 2/29/2012, so it would be a good idea to check the specifications of whatever addmonth() function you plan on using.
回答2:
It's even more crippled than that - subtraction has same issues. There are special methods to avoid overflows, though:
function original(){
return new Carbon('2016-08-31');
};
function print_dt($name, $date){
echo $name . $date->toAtomString() . PHP_EOL;
};
print_dt('original: ', original());
echo '-----' . PHP_EOL;
print_dt('addMonths: ', original()->addMonths(6));
print_dt('addMonthsNoOverflow: ', original()->addMonthsNoOverflow(6));
echo '-----' . PHP_EOL;
print_dt('subMonths: ', original()->subMonths(2));
print_dt('subMonthsNoOverflow: ', original()->subMonthsNoOverflow(2));
output:
original: 2016-08-31T00:00:00+00:00
-----
addMonths: 2017-03-03T00:00:00+00:00
addMonthsNoOverflow: 2017-02-28T00:00:00+00:00
-----
subMonths: 2016-07-01T00:00:00+00:00
subMonthsNoOverflow: 2016-06-30T00:00:00+00:00
来源:https://stackoverflow.com/questions/39207798/carbon-why-addmonths-change-the-day-of-month