Re: Unnecessary overhead in getdate()

On 29.03.2018 at 11:51, Benjamin Coutu wrote: > Just like [i]date(), getdate() > without the $timestamp argument uses the current timestamp. > > The code for [i]date() > (https://github.com/php/php-src/blob/master/ext/date/php_date.c#L1276) only makes a system call to > time(NULL) when the $timestamp > argument is missing. In contrast getdate() > (https://github.com/php/php-src/blob/master/ext/date/php_date.c#L1828) always performs a system call > to time(NULL) even if $timestamp is > supplied by the user. > > The code for getdate() should mimic what is done for > date() to avoid unnecessary system call overhead, hence only call > time(NULL) when !ZEND_NUM_ARGS(). ACK. I wonder, though, whether it is safe to convert a time_t (which is supposed to be returned by time()) to zend_long. POSIX 6 claims that time_t shall be an integer or real-floating type[1]; POSIX 7 is more specific ("time_t shall be an integer type")[2], but obviously there are neither guarantees regarding signess nor size, and it doesn't seem that we have respective checks in place[3]. [1] <http://pubs.opengroup.org/onlinepubs/009695299/basedefs/sys/types.h.html> [2] <http://pubs.opengroup.org/onlinepubs/9699919799/basedefs/sys_types.h.html> [3] <https://github.com/php/php-src/blob/master/configure.ac> -- Christoph M. Becker