Re: [PHP-DEV] Improve PHP 7 serialization
On Sep 23, 2014 8:07 PM, Leigh lei...@gmail.com wrote: On 23 September 2014 17:36, Park Framework park.framew...@gmail.com wrote: If you do not update in PHP 7 serialization method, it will never be updated, the default serialization in PHP 7 will be slow. To maintain backward compatibility, can implement support method calls on primitive types, new algorithms for serialization to be called only in the new API. $var-serialize() $var-unserialize() What do you think about this? To maintain backward compatibility we leave the behaviour of un/serialize() in tact. There are extensions for alternative serialisation methods and regardless of whether any of these becomes a bundled extension, the functionality should exist under a different set of function names. They all exist already. Either for automatic serialization or using extname_(un)serialize, or pack/unpack. Also as it has been said already, many different extensions exist, each of them being good in one or many areas. It is like compression methods, use the right one for the right task.
Re: [PHP-DEV] Improve PHP 7 serialization
Hi person hiding behind a project, Backwards compatibility is one hurdle, but if you wipe all your serialised data then begin to re-serialise using the new approach then you're fine. As for what to use msgpack or igbinary, well there's already good support for igbinary in PHP thanks to Pierre and others. You should benchmark igbinary vs msgpack and come back with your findings if we're to evaluate alternative serialization libraries. Many thanks, Paul On Tue, Sep 23, 2014 at 12:23 AM, Park Framework park.framew...@gmail.com wrote: PHP serialization is slowest in PHP Session, clients NoSQL, ... I would like to have in PHP 7, a new serialization algorithm or custom handler to serialize. My opinion is that the best choice is to use msgpack, it is +110% faster -30% data size HHVM discussed this issue, but all boils down to backward compatibility with PHP https://github.com/facebook/hhvm/issues/2654 What do you think about this, maybe it's time to change the old algorithm serialization, on something better? -- PHP Internals - PHP Runtime Development Mailing List To unsubscribe, visit: http://www.php.net/unsub.php
Re: [PHP-DEV] Improve PHP 7 serialization
Performance testing, Msgpack VS Igbinary igbinary: -20% slower, data size ~5% Advantage Msgpack, he works fast, and this format understood by many technologies - Java, Python, Lua in Redis. 2014-09-23 12:20 GMT+03:00 Paul Dragoonis dragoo...@gmail.com: Hi person hiding behind a project, Backwards compatibility is one hurdle, but if you wipe all your serialised data then begin to re-serialise using the new approach then you're fine. As for what to use msgpack or igbinary, well there's already good support for igbinary in PHP thanks to Pierre and others. You should benchmark igbinary vs msgpack and come back with your findings if we're to evaluate alternative serialization libraries. Many thanks, Paul On Tue, Sep 23, 2014 at 12:23 AM, Park Framework park.framew...@gmail.com wrote: PHP serialization is slowest in PHP Session, clients NoSQL, ... I would like to have in PHP 7, a new serialization algorithm or custom handler to serialize. My opinion is that the best choice is to use msgpack, it is +110% faster -30% data size HHVM discussed this issue, but all boils down to backward compatibility with PHP https://github.com/facebook/hhvm/issues/2654 What do you think about this, maybe it's time to change the old algorithm serialization, on something better? -- PHP Internals - PHP Runtime Development Mailing List To unsubscribe, visit: http://www.php.net/unsub.php -- PHP Internals - PHP Runtime Development Mailing List To unsubscribe, visit: http://www.php.net/unsub.php
Re: [PHP-DEV] Improve PHP 7 serialization
Write an extension for it then, also share your benchmarks :) On Tue, Sep 23, 2014 at 12:17 PM, Park Framework park.framew...@gmail.com wrote: Performance testing, Msgpack VS Igbinary igbinary: -20% slower, data size ~5% Advantage Msgpack, he works fast, and this format understood by many technologies - Java, Python, Lua in Redis. 2014-09-23 12:20 GMT+03:00 Paul Dragoonis dragoo...@gmail.com: Hi person hiding behind a project, Backwards compatibility is one hurdle, but if you wipe all your serialised data then begin to re-serialise using the new approach then you're fine. As for what to use msgpack or igbinary, well there's already good support for igbinary in PHP thanks to Pierre and others. You should benchmark igbinary vs msgpack and come back with your findings if we're to evaluate alternative serialization libraries. Many thanks, Paul On Tue, Sep 23, 2014 at 12:23 AM, Park Framework park.framew...@gmail.com wrote: PHP serialization is slowest in PHP Session, clients NoSQL, ... I would like to have in PHP 7, a new serialization algorithm or custom handler to serialize. My opinion is that the best choice is to use msgpack, it is +110% faster -30% data size HHVM discussed this issue, but all boils down to backward compatibility with PHP https://github.com/facebook/hhvm/issues/2654 What do you think about this, maybe it's time to change the old algorithm serialization, on something better? -- PHP Internals - PHP Runtime Development Mailing List To unsubscribe, visit: http://www.php.net/unsub.php
Re: [PHP-DEV] Improve PHP 7 serialization
On 23 September 2014 12:22, Paul Dragoonis dragoo...@gmail.com wrote: Write an extension for it then, also share your benchmarks :) Why go to all that trouble, 10 seconds on Google and we have: https://github.com/msgpack/msgpack-php -- PHP Internals - PHP Runtime Development Mailing List To unsubscribe, visit: http://www.php.net/unsub.php
Re: [PHP-DEV] Improve PHP 7 serialization
http://pecl.php.net/package/msgpack On Tue, Sep 23, 2014 at 1:34 PM, Leigh lei...@gmail.com wrote: On 23 September 2014 12:22, Paul Dragoonis dragoo...@gmail.com wrote: Write an extension for it then, also share your benchmarks :) Why go to all that trouble, 10 seconds on Google and we have: https://github.com/msgpack/msgpack-php -- PHP Internals - PHP Runtime Development Mailing List To unsubscribe, visit: http://www.php.net/unsub.php -- Pierre @pierrejoye | http://www.libgd.org -- PHP Internals - PHP Runtime Development Mailing List To unsubscribe, visit: http://www.php.net/unsub.php
Re: [PHP-DEV] Improve PHP 7 serialization
I clearly didn't google, it would be interesting to see comparisons of high speed PHP serialization libraries. I for one would be happy, in PHP 7, to break BC serialization syntax in favour of putting in a much faster serializer by default. Similar scenario to putting in Zend OpCache by default instead of APC. Pierre, do you see merit on including insert best overall serializer lib here by default in PHP7 ? On Tue, Sep 23, 2014 at 12:57 PM, Pierre Joye pierre@gmail.com wrote: http://pecl.php.net/package/msgpack On Tue, Sep 23, 2014 at 1:34 PM, Leigh lei...@gmail.com wrote: On 23 September 2014 12:22, Paul Dragoonis dragoo...@gmail.com wrote: Write an extension for it then, also share your benchmarks :) Why go to all that trouble, 10 seconds on Google and we have: https://github.com/msgpack/msgpack-php -- PHP Internals - PHP Runtime Development Mailing List To unsubscribe, visit: http://www.php.net/unsub.php -- Pierre @pierrejoye | http://www.libgd.org
Re: [PHP-DEV] Improve PHP 7 serialization
Park Framework wrote (on 23/09/2014): PHP serialization is slowest in PHP Session, clients NoSQL, ... I would like to have in PHP 7, a new serialization algorithm or custom handler to serialize. My opinion is that the best choice is to use msgpack, it is +110% faster -30% data size HHVM discussed this issue, but all boils down to backward compatibility with PHP https://github.com/facebook/hhvm/issues/2654 What do you think about this, maybe it's time to change the old algorithm serialization, on something better? Apart from the BC implications, using a binary serialization by default might cause issues with anyone who is storing or passing the serialized data somewhere which is not binary-safe. Admittedly, any object with private properties generates a serialized form with null bytes, but many values will consist entirely of ASCII characters, and some code may rely on this being the case. The format is also widely known, and has been implemented in other languages for compatibility (although it is **not** suitable for untrusted data exchange, as Anthony Ferrara tweeted a few months ago: https://twitter.com/ircmaxell/status/452182852562862080) We already have pluggable serializers for sessions (needed because the serialization happens implicitly in the session handling code), and can add as many functions for types of serialization as seem sensible, so I'm not sure what the benefit of changing serialize()/unserialize() themselves is. Changing the default *session* serialization might be worth considering, though, along with bundling something like igbinary or msgpack. Oh, and a non-batshit version of session_decode() for manually invoking session (un)serialization handlers :P -- Rowan Collins [IMSoP] -- PHP Internals - PHP Runtime Development Mailing List To unsubscribe, visit: http://www.php.net/unsub.php
Re: [PHP-DEV] Improve PHP 7 serialization
On Sep 23, 2014 3:31 PM, Paul Dragoonis dragoo...@gmail.com wrote: I clearly didn't google, it would be interesting to see comparisons of high speed PHP serialization libraries. I for one would be happy, in PHP 7, to break BC serialization syntax in favour of putting in a much faster serializer by default. Similar scenario to putting in Zend OpCache by default instead of APC. Pierre, do you see merit on including insert best overall serializer lib here by default in PHP7 ? Not really, not because it is not good but because there is always be a better one. We can't break format in every release.
Re: [PHP-DEV] Improve PHP 7 serialization
Not really, not because it is not good but because there is always be a better one. We can't break format in every release. If you do not update in PHP 7 serialization method, it will never be updated, the default serialization in PHP 7 will be slow. To maintain backward compatibility, can implement support method calls on primitive types, new algorithms for serialization to be called only in the new API. $var-serialize() $var-unserialize() What do you think about this? -- PHP Internals - PHP Runtime Development Mailing List To unsubscribe, visit: http://www.php.net/unsub.php
Re: [PHP-DEV] Improve PHP 7 serialization
On Tue, Sep 23, 2014 at 7:36 PM, Park Framework park.framew...@gmail.com wrote: Not really, not because it is not good but because there is always be a better one. We can't break format in every release. If you do not update in PHP 7 serialization method, it will never be updated, the default serialization in PHP 7 will be slow. To maintain backward compatibility, can implement support method calls on primitive types, new algorithms for serialization to be called only in the new API. $var-serialize() $var-unserialize() What do you think about this? Not changing the serialize() format doesn't mean that other formats can't be introduced via extensions. Though, I too would like to have more of them available by default. Cheers, Andrey. -- PHP Internals - PHP Runtime Development Mailing List To unsubscribe, visit: http://www.php.net/unsub.php
Re: [PHP-DEV] Improve PHP 7 serialization
On 23 September 2014 17:36, Park Framework park.framew...@gmail.com wrote: If you do not update in PHP 7 serialization method, it will never be updated, the default serialization in PHP 7 will be slow. To maintain backward compatibility, can implement support method calls on primitive types, new algorithms for serialization to be called only in the new API. $var-serialize() $var-unserialize() What do you think about this? To maintain backward compatibility we leave the behaviour of un/serialize() in tact. There are extensions for alternative serialisation methods and regardless of whether any of these becomes a bundled extension, the functionality should exist under a different set of function names. -- PHP Internals - PHP Runtime Development Mailing List To unsubscribe, visit: http://www.php.net/unsub.php
Re: [PHP-DEV] Improve PHP 7 serialization
Hi! I clearly didn't google, it would be interesting to see comparisons of high speed PHP serialization libraries. I for one would be happy, in PHP 7, to break BC serialization syntax in favour of putting in a much faster serializer by default. Similar scenario to putting in Zend OpCache by default instead of APC. Why break anything? If you need faster serializer, it's quite easy to get one, including msgpack. If it is really an issue that is important for people, we could include the package into core. But I don't see breaking BC in serialize/unserialize as a big win here. If it's really a bottleneck, a userspace package abstracting the specific serializer function could be easily created - and most clients like sessions already allow to switch serializers by configs. So BC break does not seem to be warranted here. -- Stanislav Malyshev, Software Architect SugarCRM: http://www.sugarcrm.com/ -- PHP Internals - PHP Runtime Development Mailing List To unsubscribe, visit: http://www.php.net/unsub.php
Re: [PHP-DEV] Improve PHP 7 serialization
On 23 September 2014 12:17:35 GMT+01:00, Park Framework park.framew...@gmail.com wrote: Performance testing, Msgpack VS Igbinary igbinary: -20% slower, data size ~5% As with any benchmark, the details of the test are rather important. Firstly, some data structures may be better handled than others, or be targeted for extra optimization, making some scenarios favour one method or the other. Some care needs to be taken in simulating one or more realistic use-cases. Secondly, speed to serialize and speed to unserialize are separate measures: igbinary openly admits that it is best used for things like caching, where reading occurs more often than writing, as it is often slower than text-based methods at write-time, but faster at read-time. Thirdly, the algorithms may have optional features which trade speed for space, or affect the above two points. For instance, igbinary's string interning, or the choice of structure used for objects in a PHP msgpack implementation. All that taken into account, it's unlikely that any one format is better in all situations, and in some cases the existing text-based format may even have measurable advantages. Which points again to the idea of making more algorithms available as bundled extensions, and as session serialization methods, but not changing the meaning of existing functions. -- Rowan Collins [IMSoP] -- PHP Internals - PHP Runtime Development Mailing List To unsubscribe, visit: http://www.php.net/unsub.php
Re: [PHP-DEV] Improve PHP 7 serialization
Why break anything? If you need faster serializer, it's quite easy to get one, including msgpack. If it is really an issue that is important for people, we could include the package into core. But I don't see breaking BC in serialize/unserialize as a big win here. If it's really a bottleneck, a userspace package abstracting the specific serializer function could be easily created - and most clients like sessions already allow to switch serializers by configs. So BC break does not seem to be warranted here. Perhaps a compromise would be to choose the quickest method of serialization, add it to PHP core. In php.ini add the directive serialization.method = msgpack / Igbinary / -- PHP Internals - PHP Runtime Development Mailing List To unsubscribe, visit: http://www.php.net/unsub.php
Re: [PHP-DEV] Improve PHP 7 serialization
Hi! Perhaps a compromise would be to choose the quickest method of serialization, add it to PHP core. In php.ini add the directive serialization.method = msgpack / Igbinary / We could, but what if you need to read/write data specifically from current PHP serializer? You'd have to mess then with runtime directives, it doesn't look like a good design. That's like having one db_query() function for all databases and have a config parameter that switches the global database type. I think the other option is better - to have extensions for all underlying functions and abstraction layer (PDO or userspace) that provides unified API if needed. -- Stanislav Malyshev, Software Architect SugarCRM: http://www.sugarcrm.com/ -- PHP Internals - PHP Runtime Development Mailing List To unsubscribe, visit: http://www.php.net/unsub.php
Re: [PHP-DEV] Improve PHP 7 serialization
I agree, you're right. My desire to override the existing algorithm serialize(), due to the need to change the method serialization, but does not change the source code (legacy code, ext PHP) 2014-09-24 3:03 GMT+03:00 Stas Malyshev smalys...@sugarcrm.com: Hi! Perhaps a compromise would be to choose the quickest method of serialization, add it to PHP core. In php.ini add the directive serialization.method = msgpack / Igbinary / We could, but what if you need to read/write data specifically from current PHP serializer? You'd have to mess then with runtime directives, it doesn't look like a good design. That's like having one db_query() function for all databases and have a config parameter that switches the global database type. I think the other option is better - to have extensions for all underlying functions and abstraction layer (PDO or userspace) that provides unified API if needed. -- Stanislav Malyshev, Software Architect SugarCRM: http://www.sugarcrm.com/ -- PHP Internals - PHP Runtime Development Mailing List To unsubscribe, visit: http://www.php.net/unsub.php
Re: [PHP-DEV] Improve PHP 7 serialization
Hi 2014-09-23 23:56 GMT+02:00 Park Framework park.framew...@gmail.com: In php.ini add the directive serialization.method = msgpack / Igbinary / There is an even better way to do this; add an additional parameter to serialize and unserialize to serialize as and unserialize as: $bin = serialize($data_struct, 'igbinary'); $data_struct = unserialize($bin, 'igbinary'); This keeps a clean BC solution without adding more clutter to the ini file for runtime things, you could say that adding something like 'serialize_default_mode = php;', but changing its default would create clutter code where the optional second parameter would be forced, so lets leave the ini out of this. So what I propose here is: 1) An internal API to register serializes,and making PHP's current serialize implemented as such 2) Add a second parameter to both serialize and unserialize that can be used choose a serializer 3) Optionally add a function like: get_serialize_handlers() (so we won't have to parse phpinfo()) which are the available serializers 4) Consider bundling either igbinary, msgpack or implement a new custom and more efficient one that allows us to be future oriented 5) No php.ini changes 6) Allow users to register serialize handlers using register_serialize_handler()/unregister_serialize_handler() 7) Optionally consider implementing this in SPL Okay, late night, back to sleep! -- regards, Kalle Sommer Nielsen ka...@php.net -- PHP Internals - PHP Runtime Development Mailing List To unsubscribe, visit: http://www.php.net/unsub.php
Re: [PHP-DEV] Improve PHP 7 serialization
Hi! There is an even better way to do this; add an additional parameter to serialize and unserialize to serialize as and unserialize as: $bin = serialize($data_struct, 'igbinary'); $data_struct = unserialize($bin, 'igbinary'); This is cleaner, but if you can do this (code change), why you can't do just igbinary_serialize($data_struct)? 3) Optionally add a function like: get_serialize_handlers() (so we won't have to parse phpinfo()) which are the available serializers That actually makes a lot of sense, but serialize_get_handlers() might be a better name, to group them together. But right now I don't think we have such list, do we? We have php_session_register_serializer() and the list for sessions, but not for other contexts. 6) Allow users to register serialize handlers using register_serialize_handler()/unregister_serialize_handler() Do you think userspace serialize handlers would be popular? They would be by necessity pretty slow compared to C ones. -- Stanislav Malyshev, Software Architect SugarCRM: http://www.sugarcrm.com/ -- PHP Internals - PHP Runtime Development Mailing List To unsubscribe, visit: http://www.php.net/unsub.php
[PHP-DEV] Improve PHP 7 serialization
PHP serialization is slowest in PHP Session, clients NoSQL, ... I would like to have in PHP 7, a new serialization algorithm or custom handler to serialize. My opinion is that the best choice is to use msgpack, it is +110% faster -30% data size HHVM discussed this issue, but all boils down to backward compatibility with PHP https://github.com/facebook/hhvm/issues/2654 What do you think about this, maybe it's time to change the old algorithm serialization, on something better? -- PHP Internals - PHP Runtime Development Mailing List To unsubscribe, visit: http://www.php.net/unsub.php
Re: [PHP-DEV] Improve PHP 7 serialization
On Sep 23, 2014 1:23 AM, Park Framework park.framew...@gmail.com wrote: PHP serialization is slowest in PHP Session, clients NoSQL, ... I would like to have in PHP 7, a new serialization algorithm or custom handler to serialize. The latter is already possible and there are many good extensions doing that. Igbinary f.e. My opinion is that the best choice is to use msgpack, it is +110% faster -30% data size HHVM discussed this issue, but all boils down to backward compatibility with PHP https://github.com/facebook/hhvm/issues/2654 What do you think about this, maybe it's time to change the old algorithm serialization, on something better? -- PHP Internals - PHP Runtime Development Mailing List To unsubscribe, visit: http://www.php.net/unsub.php
