|
php.net | support | documentation | report a bug | advanced search | search howto | statistics | random bug | login |
[2019-09-27 14:04 UTC] chris dot smith at widerplan dot com
Description:
------------
I'm not sure if this is a problem with the documentation or code.
The documentation states:
> If format does not contain the character ! then portions of the generated time which are not specified in format will be set to the current system time.
When executing the following based on the documentation I'd expect the hour, minute and seconds to be set to the current time however they are set to the epoch.
DateTime::createFromFormat('u', '123456');
// PHP < 7.3: 2019-09-27 14:50:40.123456
// PHP >= 7.3: 2019-09-27 00:00:00.123456
Further investigation led me to discover that the following example also contradicts the documentation and has been the case since the days of PHP 5.3.
DateTimeImmutable::createFromFormat('s', '12');
// 2019-09-27 00:00:12.000000
// Expected (per docs): 2019-09-27 14:50:12.000000
https://3v4l.org/v2bib
I ran a git bisect to identify when the microsecond handling changed which found: 09340ea9b9b2a06c6f1369502806640779073885
Test script:
---------------
<?php
$dt1 = DateTime::createFromFormat('u', '123456');
$dt2 = DateTime::createFromFormat('s', '25');
var_dump($dt1->format('H:i'));
var_dump($dt2->format('H:i:s'));
Expected result:
----------------
string(5) "14:29"
string(8) "14:29:25"
Actual result:
--------------
string(5) "00:00"
string(8) "00:00:25"
PatchesPull RequestsHistoryAllCommentsChangesGit/SVN commits
|
|||||||||||||||||||||||||||||||||||||
Copyright © 2001-2025 The PHP GroupAll rights reserved. |
Last updated: Fri Oct 24 06:00:01 2025 UTC |
> Testing it against anything pre 7.3 and it works as the documentation shows. If you do DateTime::createFromFormat('s', '12'); then the other parts of the time component are set to zero, i.e. the epoch this has been the same since 5.3 which does not follow the documentation. As of 7.3 `u` now works the same as seconds/minutes/hours, question is which behaviour is correct. My thinking is that it should do what the documentation says but it's been broken so long and nobody has noticed it, that it makes more sense to change the documentation to match reality.