php-src/UPGRADING
Niels Dossche 7658220599
Improve performance of mbfl_name2encoding() by using perfect hashing (#12707)
mbfl_name2encoding() uses a linear loop through the encodings, comparing
the name one by one, which is very slow. For the benchmark [1] just
looking up the name takes about 50% of run-time.

By using perfect hashing instead, we no longer have to loop over the
list, and the number of string comparisons is reduced to just a single
one. The perfect hashing table is generated using GNU gperf and amended
manually to fit in with mbstring and manually changed to  reduce the
cache size.

[1] https://github.com/php/php-src/issues/12684#issuecomment-1813799924
2023-11-17 19:38:43 +01:00

232 lines
8.4 KiB
Plaintext

PHP 8.4 UPGRADE NOTES
1. Backward Incompatible Changes
2. New Features
3. Changes in SAPI modules
4. Deprecated Functionality
5. Changed Functions
6. New Functions
7. New Classes and Interfaces
8. Removed Extensions and SAPIs
9. Other Changes to Extensions
10. New Global Constants
11. Changes to INI File Handling
12. Windows Support
13. Other Changes
14. Performance Improvements
========================================
1. Backward Incompatible Changes
========================================
- DOM:
. New methods and constants were added to some DOM classes. If you inherit
from these and you happen to have a method or property with the same name,
you might encounter errors if the declaration is incompatible.
Consult sections 2. New Features and 6. New Functions for a list of
newly implemented methods and constants.
- PDO_PGSQL:
. The DSN's credentials, when set, are given priority over their PDO
constructor counterparts, being closer to the documentation states.
- PCNTL:
. The functions pcntl_sigprocmask(), pcntl_sigwaitinfo() and
pcntl_sigtimedwait() now throw:
- A ValueError if the $signals array is empty (except for
pcntl_sigprocmask() if the $mode is SIG_SETMASK).
- A TypeError if a value of the $signals array is not an integer
- A ValueError if a value of the $signals array is not a valid signal number
Moreover, those functions now always return false on failure.
In some case previously it could return the value -1.
. The function pcntl_sigprocmask() will also now throw:
- A ValueError if $mode is not one of SIG_BLOCK, SIG_UNBLOCK, or SIG_SETMASK
. The function pcntl_sigtimedwait() will also now throw:
- A ValueError if $seconds is less than 0
- A ValueError if $nanoseconds is less than 0 or greater than 1e9
- A ValueError if both $seconds and $nanoseconds are 0
- SimpleXML:
. Get methods called, or casting to a string on a SimpleXMLElement will no
longer implicitly reset the iterator data, unless explicitly rewound.
For example, casting an element to a string within a foreach loop would
cause an infinite loop because it destroyed the current iterator data.
This is no longer the case as a consequence of the bugfixes for GH-12192,
GH-12208, #55098.
- SPL:
. Out of bounds accesses in SplFixedArray now throw an exception of type
OutOfBoundsException instead of RuntimeException. As OutOfBoundsException
is a child class of RuntimeException, code that uses RuntimeException
continues to function.
- Standard:
. round() now validates the value of the $mode parameter and throws a ValueError
for invalid modes. Previously invalid modes would have been interpreted as
PHP_ROUND_HALF_UP.
. strcspn() with empty $characters now returns the length of the string instead
of incorrectly stopping at the first NUL character. See GH-12592.
- XML:
. The xml_set_*_handler() functions now declare and check for an effective
signature of callable|string|null for the $handler parameters.
Moreover, values of type string that correspond to method names,
of object set with xml_set_object() are now checked to see if the method
exists on the class of the previously passed object.
This means that xml_set_object() must now always be called prior to setting
method names as callables.
Passing an empty string to disable the handler is still allowed,
but not recommended.
- XSL:
. XSLTProcessor::setParameter() will now throw a ValueError when its arguments
contain null bytes. This never actually worked correctly in the first place,
which is why it throws an exception nowadays.
. The typed properties XSLTProcessor::$cloneDocument and
XSLTProcessor::$doXInclude are now declared.
========================================
2. New Features
========================================
- DOM:
. Added constant DOMNode::DOCUMENT_POSITION_DISCONNECTED.
. Added constant DOMNode::DOCUMENT_POSITION_PRECEDING.
. Added constant DOMNode::DOCUMENT_POSITION_FOLLOWING.
. Added constant DOMNode::DOCUMENT_POSITION_CONTAINS.
. Added constant DOMNode::DOCUMENT_POSITION_CONTAINED_BY.
. Added constant DOMNode::DOCUMENT_POSITION_IMPLEMENTATION_SPECIFIC.
. Implemented DOM HTML5 parsing and serialization.
RFC: https://wiki.php.net/rfc/domdocument_html5_parser.
This RFC adds the new DOM namespace along with class and constant aliases.
There are two new classes to handle HTML and XML documents:
DOM\HTMLDocument and DOM\XMLDocument.
These classes provide a cleaner API to handle HTML and XML documents.
Furthermore, the DOM\HTMLDocument class implements spec-compliant HTML5
parsing and serialization.
- Phar:
. Added support for the unix timestamp extension for zip archives.
- POSIX:
. Added constant POSIX_SC_CHILD_MAX
. Added constant POSIX_SC_CLK_TCK
- SOAP:
. Added support for clark notation for namespaces in class map.
It is now possible to specify entries in a class map with clark notation
to resolve a type with a specific namespace to a specific class.
For example: '{http://example.com}foo' => 'FooClass'.
- XSL:
. It is now possible to use parameters that contain both single and double
quotes.
========================================
3. Changes in SAPI modules
========================================
========================================
4. Deprecated Functionality
========================================
========================================
5. Changed Functions
========================================
- MBString:
. The behavior of mb_strcut is more consistent now on invalid UTF-8 and UTF-16
strings. (For valid UTF-8 and UTF-16 strings, there is no change.)
- PGSQL:
. pg_select, the conditions arguments accepts an empty array and is optional.
- Standard:
. The internal implementation for rounding to integers has been rewritten
to be easier to verify for correctness and to be easier to maintain.
Some rounding bugs have been fixed as a result of the rewrite. For
example previously rounding 0.49999999999999994 to the nearest integer
would have resulted in 1.0 instead of the correct result 0.0. Additional
inputs might also be affected and result in different outputs compared to
earlier PHP versions.
. The default value of the 'cost' option for PASSWORD_BCRYPT for password_hash()
has been increased from '10' to '12'.
RFC: https://wiki.php.net/rfc/bcrypt_cost_2023
========================================
6. New Functions
========================================
- DOM:
. Added DOMNode::compareDocumentPosition().
========================================
7. New Classes and Interfaces
========================================
========================================
8. Removed Extensions and SAPIs
========================================
========================================
9. Other Changes to Extensions
========================================
- Intl:
. The class constants are typed now.
- PDO:
. The class constants are typed now.
- Reflection:
. The class constants are typed now.
- Spl:
. The class constants are typed now.
- Sqlite:
. The class constants are typed now.
========================================
10. New Global Constants
========================================
- Intl:
. The IntlDateFormatter class exposes now the new PATTERN constant
reflecting udat api's UDAT_PATTERN.
- XML:
. Added XML_OPTION_PARSE_HUGE to allow large inputs in xml_parse and
xml_parse_into_struct.
RFC: https://wiki.php.net/rfc/xml_option_parse_huge.
========================================
11. Changes to INI File Handling
========================================
========================================
12. Windows Support
========================================
========================================
13. Other Changes
========================================
========================================
14. Performance Improvements
========================================
* The performance of DOMNode::C14N() is greatly improved for the case without
an xpath query. This can give a time improvement of easily two order of
magnitude for documents with tens of thousands of nodes.
* The performance of strspn() and strcspn() is greatly improved.
They now run in linear time instead of being bounded by quadratic time.
* mb_strcut() is much faster now for UTF-8 and UTF-16 strings.
* get_browser() is much faster now, up to 1.5x - 2.5x for some test cases.
* Looking up mbstring encoding names is much faster now.