No need for CLS_FETCH().

Fix broken ZTS build

Modified /F handling to allow calling object methods.

If we can't call custom replacement function, don't change the subject
string.

Don't need to check for out-of-memory condition.

Z-ify the source.

Fix argument check (bug #8421).

(PHP preg_replace) Fix to allow proper escaping of captured subpattern
references in the replacement string.
(PHP array_sum) Check that argument is an array.

Heads up people!

Updated the get_current_key() API - the relevant authors, please take
a look at the updated code and make sure it's ok...

Oh man, gcc's logic is really flawed

Kill a warning (code is correct)

Fix the arg array size.

This submission introduces two features.

The first one is support for Perl-style matching regexp delimiters, i.e.
using <[{( and )}]> to delimit the regular expressions.

The sec

This submission introduces two features.

The first one is support for Perl-style matching regexp delimiters, i.e.
using <[{( and )}]> to delimit the regular expressions.

The second one is a new 'F' modifier that allows you to specify a function name
in the replacement argument to preg_replace(). This function will be called
when the replacement needs to be made. It is passed an array of full matched
pattern and captured subpatterns and it is expected to return a string that
will be used for replacement. 'e' and 'F' modifiers cannot be used together.

@- Implemented support for Perl-style matching regexp delimiters in PCRE.
@ You can use <{[( and )]}> to delimit your expressions now. (Andrei)

@- Introduced new 'F' modifier in PCRE that lets you specify a function
@ name in the replacement argument to preg_replace() that will be called
@ at run-time to provide the replacement string. It is passed an array of
@ matched pattern and subpatterns. (Andrei)

show more ...

- Modified preg_replace() to take Perl-style $1 backreferences in addition
to \\n ones.
- Modified preg_grep() to always return the results with the keys
from the original array.

@- Modified preg_replace() to ignore backreferences that refer to
@ non-existing subpatterns. (Andrei)

Fixed bug #6740.
I happen to think that this is php_addslashes() problem, not PCRE's.
When 0 is passed for the length of the string to php_addslashes() it
assumes that we want to process

Fixed bug #6740.
I happen to think that this is php_addslashes() problem, not PCRE's.
When 0 is passed for the length of the string to php_addslashes() it
assumes that we want to process the whole string and happily runs
strlen() on it. That is bad. It should respect the length and return
an empty string if it's 0.

show more ...

- Made eval() and several other runtime-evaluated code portions report the
nature and location of errors more accurately (Stas)

@- Fixed a bug in preg_replace() that would truncate subject string if the
@ first argument was en empty array. (Andrei)
This fixes bug #6346, that would truncate subject string if the
f

@- Fixed a bug in preg_replace() that would truncate subject string if the
@ first argument was en empty array. (Andrei)
This fixes bug #6346, that would truncate subject string if the
first argument was en empty array.

show more ...

@- Fixed a bug that did not respect limit parameter in preg_replace() when
@ using /e modifier. (Andrei)

Reverting the pemalloc -> malloc change until we figure out why it breaks.

No need to have persistent allocation.

# There you go, Wico.
Fix bug $5676.
@- Fixed preg_replace() to automatically escape quotes in matched
@ strings when using /e modifier. (Andrei)

Typo.

@- Added an optional parameter to preg_replace() that can be used to
@ specify how many replacements to make. (Andrei)

@- Upgraded PCRE to version 3.2 and fixed a bug when anchored pattern
@ matched an empty string. (Andrei)