ID: 50075 User updated by: RQuadling at GMail dot com Reported By: RQuadling at GMail dot com Status: Open Bug Type: Feature/Change Request Operating System: Irrelevant PHP Version: 5.3.1RC3 New Comment:
Which would suggest to me that DateTime->diff() has a bug. Personally, if I'm asking for the difference between 2 dates, the physical time between them would be what I would expect to be reported. So, when checking over DST, the result could be 23 hours (GMT->BST) or 1 day + 1 hour (BST -> GMT). Neither are 1 day. Previous Comments: ------------------------------------------------------------------------ [2009-11-04 15:56:14] der...@php.net DST is relevant because the following will show "one day", but it is not 86400 seconds: <?php date_default_timezone_set( "Europe/London" ); $a = new DateTime( "2009-10-25 00:00" ); $b = new DateTime( "2009-10-26 00:00" ); $d = $a->diff( $b ); echo $d->format( '%d %h:%i:%s' ), "\n"; echo $b->format( 'U' ) - $a->format( 'U' ), "\n"; echo $a->format( DateTime::ISO8601 ), "\n"; echo $b->format( DateTime::ISO8601 ), "\n"; ?> ------------------------------------------------------------------------ [2009-11-04 15:43:34] RQuadling at GMail dot com I'm not sure how DST is relevant here. DateInterval will already have worked out the difference in days, hours, mins, etc. All I'm providing is a total number of seconds. I take on the issue with invert, but it does mean having to do some userland calcs. diff -> seconds. Yep. Obviously. Index: php_date.c =================================================================== --- php_date.c (revision 290198) +++ php_date.c (working copy) @@ -2245,6 +2245,15 @@ PHP_DATE_INTERVAL_ADD_PROPERTY("s", s); PHP_DATE_INTERVAL_ADD_PROPERTY("invert", invert); PHP_DATE_INTERVAL_ADD_PROPERTY("days", days); + + MAKE_STD_ZVAL(zv); + ZVAL_LONG(zv, ((((( + intervalobj->diff->days * 24) + + intervalobj->diff->h) * 60) + + intervalobj->diff->i) * 60) + + intervalobj->diff->s + ); + zend_hash_update(props, "seconds", 8, &zv, sizeof(zval), NULL); return props; } ------------------------------------------------------------------------ [2009-11-04 15:09:16] der...@php.net This patch is not correct, because it doesn't take care of daylight savings time nor should "invert" being taken into account as it's already a seperate field. I wouldn't call it "diff" either, but "seconds" just like there is "days". ------------------------------------------------------------------------ [2009-11-04 14:31:46] RQuadling at GMail dot com Description: ------------ The output of DateInterval->diff() currently shows the breakdown of the difference (y, m, d, h, i, s, days, invert). Which is fine. But I'd like to be able to have the number of seconds difference. The following patch adds a new entry to the output: diff. It is a straight calculation of the seconds based upon days, h, i and s, and takes into consideration invert. The patch below is the same for branch/5_3 and trunk. I think this patch and the DateInterval object comparison patch reported in Bug#49914 by aharvey would be good additions. Regards, Richard. Reproduce code: --------------- Index: php_date.c =================================================================== --- php_date.c (revision 290198) +++ php_date.c (working copy) @@ -2245,6 +2245,16 @@ PHP_DATE_INTERVAL_ADD_PROPERTY("s", s); PHP_DATE_INTERVAL_ADD_PROPERTY("invert", invert); PHP_DATE_INTERVAL_ADD_PROPERTY("days", days); + + MAKE_STD_ZVAL(zv); + ZVAL_LONG(zv, (((((( + intervalobj->diff->days * 24) + + intervalobj->diff->h) * 60) + + intervalobj->diff->i) * 60) + + intervalobj->diff->s) * ( + intervalobj->diff->invert ? -1 : 1) + ); + zend_hash_update(props, "diff", 5, &zv, sizeof(zval), NULL); return props; } ------------------------------------------------------------------------ -- Edit this bug report at http://bugs.php.net/?id=50075&edit=1