Tim Galyean
2013-01-24 21:30:56 UTC
Hey guys,
Found something interesting here. It seems that flush works differently in PHP5.3 and I am trying to understand the change. Below is a small snippet of code:
<?php
echo 'a';
sleep(10);
?>
In php 5.2 running `$>php file.php` with this code in it, the 'a' prints almost immediately. In php-5.3.15 doing the same thing, it sleeps for 10 seconds then prints the 'a'. I ran an strace on this command and it also reflects this change. I have already verified that output buffering is disabled on both systems to rule that out. Any thoughts?
PHP 5.2:
write(1, "a", 1a) = 1
rt_sigprocmask(SIG_BLOCK, [CHLD], [], 8) = 0
rt_sigaction(SIGCHLD, NULL, {SIG_DFL, [], 0}, 8) = 0
rt_sigprocmask(SIG_SETMASK, [], NULL, 8) = 0
nanosleep({10, 0},
{10, 0}) = 0
PHP 5.3:
rt_sigprocmask(SIG_SETMASK, [], NULL, 8) = 0
nanosleep({10, 0},
{10, 0}) = 0
write(1, "a", 1a) = 1
gettimeofday({1359049071, 137281}, NULL) = 0
Found something interesting here. It seems that flush works differently in PHP5.3 and I am trying to understand the change. Below is a small snippet of code:
<?php
echo 'a';
sleep(10);
?>
In php 5.2 running `$>php file.php` with this code in it, the 'a' prints almost immediately. In php-5.3.15 doing the same thing, it sleeps for 10 seconds then prints the 'a'. I ran an strace on this command and it also reflects this change. I have already verified that output buffering is disabled on both systems to rule that out. Any thoughts?
PHP 5.2:
write(1, "a", 1a) = 1
rt_sigprocmask(SIG_BLOCK, [CHLD], [], 8) = 0
rt_sigaction(SIGCHLD, NULL, {SIG_DFL, [], 0}, 8) = 0
rt_sigprocmask(SIG_SETMASK, [], NULL, 8) = 0
nanosleep({10, 0},
{10, 0}) = 0
PHP 5.3:
rt_sigprocmask(SIG_SETMASK, [], NULL, 8) = 0
nanosleep({10, 0},
{10, 0}) = 0
write(1, "a", 1a) = 1
gettimeofday({1359049071, 137281}, NULL) = 0