[wp-trac] [WordPress Trac] #18007: is_serialized trouble for multibytes encoding

[WordPress Trac]

#18007: is_serialized trouble for multibytes encoding
----------------------------------------+------------------------------
 Reporter:  challet                     |       Owner:
     Type:  defect (bug)                |      Status:  new
 Priority:  normal                      |   Milestone:  Awaiting Review
Component:  Charset                     |     Version:  3.2
 Severity:  normal                      |  Resolution:
 Keywords:  has-patch needs-unit-tests  |
----------------------------------------+------------------------------

Comment (by SergeyBiryukov):

 I can reproduce the original bug.

 [attachment:18007.2.patch] is a more complete patch using `substr()`.

 [attachment:18007.2.alt.patch] is an another take using `count(
 str_split() )` instead of `strlen()`.

 Unfortunately, a quick test with 3 000 000 iterations of `is_serialized()`
 showed that `substr()` is about 2 times slower than direct access, and
 `count( str_split() )` is 3 times slower than `strlen()`:

 * 11,617s — unpatched
 * 24,617s — [attachment:18007.2.patch]
 * 33,553s — [attachment:18007.2.alt.patch]

 [attachment:18007.3.patch] is an attempt to only use `substr()` if
 `mbstring.func_overload` is enabled, and to keep decent performance
 otherwise:

 * 12,503s — [attachment:18007.3.patch], `mbstring.func_overload` is off
 * 26,384s — [attachment:18007.3.patch], `mbstring.func_overload` is on

 We have a similar check in `streams.php`: [[BR]]
 http://core.trac.wordpress.org/browser/tags/3.4.2/wp-
 includes/pomo/streams.php#L17

 Replying to [comment:7 kurtpayne]:
 > I'm attaching the unit test I'm working on, but I'm stuck until I get
 some feedback.  You have to apply it then run it as:
 >
 > `php -d mbstring.func_overload=4 /usr/bin/phpunit -c mbstring.xml`

 Note that `mbstring.func_overload=4` only overloads regular expression
 functions. For string functions, you need `mbstring.func_overload=2`.

 [attachment:18007.unit-test.patch] is the shortened and revised unit test.

-- 
Ticket URL: <http://core.trac.wordpress.org/ticket/18007#comment:8>
WordPress Trac <http://core.trac.wordpress.org/>
WordPress blogging software