[PATCH] Optional scalar type hinting

php.internals

Sam Barrow

18 years ago
I found a patch by Derick online to allow for scalar type hinting, and made it work with the newest snapshot of PHP 5.3. I also added the ability to type hint for resources. I would like to ask that it be added to the next PHP release. It allows type hinting for int, float, bool, string, resource, and object, I also added the ability to use the secondary keywords for all of these types (such as double, real, long, etc.). It will maintain 100% backwards compatibility, as the type hinting is 100% optional, implemented in the same way as array/class type hinting. I have the patch on my PC, please let me know where and when i can submit it. I'd be happy to do the patching and submission myself, just asking for permission here.

Richard Quadling

18 years ago
On 15/11/2007, Sam Barrow <sam@sambarrow.com> wrote:
> > I found a patch by Derick online to allow for scalar type hinting, and > made it work with the newest snapshot of PHP 5.3. I also added the > ability to type hint for resources. I would like to ask that it be added > to the next PHP release. It allows type hinting for int, float, bool, > string, resource, and object, I also added the ability to use the > secondary keywords for all of these types (such as double, real, long, > etc.). > > It will maintain 100% backwards compatibility, as the type hinting is > 100% optional, implemented in the same way as array/class type hinting. > > I have the patch on my PC, please let me know where and when i can > submit it. I'd be happy to do the patching and submission myself, just > asking for permission here.
What happens for type conversion? Is the param cast to the hinted type? The idea of type hinting for array and class is to make sure you get something appropriate. Rarely would you think of an array and cast it as an integer (or vice versa), so it makes sense. But with a string ("0"), an integer (0) or a boolean (false), they are all the same, so does this mean we would be having to cast all the params to the function/method?
-- ----- Richard Quadling Zend Certified Engineer : http://zend.com/zce.php?c=ZEND002498&r=213474731 "Standing on the shoulders of some very clever giants!"

Jeremy Privett

18 years ago
Richard Quadling wrote:
> On 15/11/2007, Sam Barrow <sam@sambarrow.com> wrote: > >> I found a patch by Derick online to allow for scalar type hinting, and >> made it work with the newest snapshot of PHP 5.3. I also added the >> ability to type hint for resources. I would like to ask that it be added >> to the next PHP release. It allows type hinting for int, float, bool, >> string, resource, and object, I also added the ability to use the >> secondary keywords for all of these types (such as double, real, long, >> etc.). >> >> It will maintain 100% backwards compatibility, as the type hinting is >> 100% optional, implemented in the same way as array/class type hinting. >> >> I have the patch on my PC, please let me know where and when i can >> submit it. I'd be happy to do the patching and submission myself, just >> asking for permission here. >> > > What happens for type conversion? Is the param cast to the hinted type? > > The idea of type hinting for array and class is to make sure you get > something appropriate. > > Rarely would you think of an array and cast it as an integer (or vice > versa), so it makes sense. > > But with a string ("0"), an integer (0) or a boolean (false), they are > all the same, so does this mean we would be having to cast all the > params to the function/method? >
I imagine that it will behave the same was as the other type hinting and just bomb out if the incoming data is of the wrong type. It kind of defeats the purpose of type "hinting" if it's really just type "casting" in the method call. To your point, I think it would really only help those OCD developers among us who always use === and try to make PHP behave like a strongly typed language. --- Jeremy Privett C.E.O. & C.S.A. Omega Vortex Corporation

Marcus Börger

18 years ago
Hello Sam, we discussed this a million times already I think. And the conclusion is to not allow type hinting for base types. marcus Thursday, November 15, 2007, 4:27:17 PM, you wrote:
> I found a patch by Derick online to allow for scalar type hinting, and > made it work with the newest snapshot of PHP 5.3. I also added the > ability to type hint for resources. I would like to ask that it be added > to the next PHP release. It allows type hinting for int, float, bool, > string, resource, and object, I also added the ability to use the > secondary keywords for all of these types (such as double, real, long, > etc.).
> It will maintain 100% backwards compatibility, as the type hinting is > 100% optional, implemented in the same way as array/class type hinting.
> I have the patch on my PC, please let me know where and when i can > submit it. I'd be happy to do the patching and submission myself, just > asking for permission here.
Best regards, Marcus

Derick Rethans

18 years ago
On Thu, 15 Nov 2007, Marcus Boerger wrote:
> we discussed this a million times already I think. And the conclusion is > to not allow type hinting for base types.
I am actually thinking that it might be a good thing to add more and more. I know my quick hack isn't the best implementation though. Derick
-- Derick Rethans http://derickrethans.nl | http://ez.no | http://xdebug.org

Cristian Rodriguez

18 years ago
2007/11/18, Derick Rethans <derick@php.net>:
> I am actually thinking that it might be a good thing to add more and > more. I know my quick hack isn't the best implementation though.
Yes and it is an alternative and not a mandatory thing to use.. as long as : <?php function foo(int $num) { return $num } foo("12345") // emit fatal error, NOT accept it as valid integer all is fine and Im all for it ;)
-- http://www.kissofjudas.net/

David Coallier

18 years ago
On Nov 18, 2007 5:52 PM, Cristian Rodriguez <judas.iscariote@gmail.com> wrote:
> 2007/11/18, Derick Rethans <derick@php.net>: > > > I am actually thinking that it might be a good thing to add more and > > more. I know my quick hack isn't the best implementation though. > > Yes and it is an alternative and not a mandatory thing to use.. as long as : > > <?php > > function foo(int $num) { > return $num > > } > > foo("12345") // emit fatal error, NOT accept it as valid integer > > all is fine and Im all for it ;) > -- > http://www.kissofjudas.net/ >
I was thinking at something along the lines of objects also for instance: $i = new Integer(33); function foo(Integer $var) { } foo ($i); else it emits a fatal error. But also if you do $i = "Name"; that would emit a fatal error because the value is suposed to be an int. This might look a bit too much like java, but as an extension it could be something quite interesting I believe. String, Object, Integer, Scalar, Float and what else. So thinking of something like $string = new String("Foo"); $string = "bar" or $string->setValue("Bar"); would do $float = new Float(4.242); $float->setValue('foobar'); // That emits an error $float->setValue(3.14159); echo $float; (__toString) or echo $float->getValue; to echo it's content/value and so on. Would that be "too" java-ish to be something considered in php6 ? ;-)
> > -- > PHP Internals - PHP Runtime Development Mailing List > To unsubscribe, visit: http://www.php.net/unsub.php > >
-- David Coallier, Founder & Software Architect, Agora Production (http://agoraproduction.com) 51.42.06.70.18

Sam Barrow

18 years ago
I have thought about this, using obects for all variables, but all of the extra code amounts to alot in a large application, and when using hundreds of small strings and integers scattered throughtout your application, this would cause a serious performance drop. On Sun, 2007-11-18 at 18:13 -0500, David Coallier wrote:

Hannes Magnusson

18 years ago
On Nov 19, 2007 12:13 AM, David Coallier <davidc@php.net> wrote:
> I was thinking at something along the lines of objects also for instance: > > $i = new Integer(33); > > function foo(Integer $var) { > } > > foo ($i); else it emits a fatal error. But also if you do > > $i = "Name"; that would emit a fatal error because the value is > suposed to be an int. This might look a bit too much like java, but as > an extension it could be something quite interesting I believe. > > String, Object, Integer, Scalar, Float and what else. > > So thinking of something like > > $string = new String("Foo"); > $string = "bar" or $string->setValue("Bar"); would do > > $float = new Float(4.242); > $float->setValue('foobar'); // That emits an error > $float->setValue(3.14159); > > echo $float; (__toString) or echo $float->getValue; to echo it's content/value > > and so on.
That has got to be the worst idea I've heard on internals for over a month. Besides, you can do this in userland already anyway: <?php class InvalidTypeException extends Exception {} class UnknownTypeException extends Exception {} class Types { const INTEGER = 1; const FLOAT = 2; const STRING = 3; const OBJECT = 4; const BOOLEAN = 5; private $val, $type; public function __construct($val, $type) { $this->type = $type; $this->setValue($val); } public function setValue($val) { switch($this->type) { case self::INTEGER: if (!is_int($val)) { throw new InvalidTypeException; } break; case self::FLOAT: if (!is_float($val)) { throw new InvalidTypeException; } break; case self::STRING: if (!is_string($val)) { throw new InvalidTypeException; } break; case self::OBJECT: if (!is_object($val)) { throw new InvalidTypeException; } break; case self::BOOLEAN: if (!is_bool($val)) { throw new InvalidTypeException; } break; default: throw new UnknownTypeException; } $this->val = $val; } public function getValue() { return $this->val; } public function __toString() { return (string)$this->getValue(); } } class Integer extends Types { public function __construct($val) { parent::__construct($val, Types::INTEGER); } } class String extends Types { public function __construct($val) { parent::__construct($val, Types::STRING); } } class Float extends Types { public function __construct($val) { parent::__construct($val, Types::FLOAT); } } class Object extends Types { public function __construct($val) { parent::__construct($val, Types::OBJECT); } } class Boolean extends Types { public function __construct($val) { parent::__construct($val, Types::BOOLEAN); } } function type_hint_integer(Integer $val) { echo $val, "\n"; } function type_hint_string(String $val) { echo $val, "\n"; } function type_hint_float(Float $val) { echo $val, "\n"; } type_hint_integer(new Integer(123)); type_hint_string(new String("string")); type_hint_float(new Float(0.25)); -Hannes

David Coallier

18 years ago
On Nov 18, 2007 7:24 PM, Hannes Magnusson <hannes.magnusson@gmail.com> wrote:
> On Nov 19, 2007 12:13 AM, David Coallier <davidc@php.net> wrote: > > I was thinking at something along the lines of objects also for instance: > > > > $i = new Integer(33); > > > > function foo(Integer $var) { > > } > > > > foo ($i); else it emits a fatal error. But also if you do > > > > $i = "Name"; that would emit a fatal error because the value is > > suposed to be an int. This might look a bit too much like java, but as > > an extension it could be something quite interesting I believe. > > > > String, Object, Integer, Scalar, Float and what else. > > > > So thinking of something like > > > > $string = new String("Foo"); > > $string = "bar" or $string->setValue("Bar"); would do > > > > $float = new Float(4.242); > > $float->setValue('foobar'); // That emits an error > > $float->setValue(3.14159); > > > > echo $float; (__toString) or echo $float->getValue; to echo it's content/value > > > > and so on. > > That has got to be the worst idea I've heard on internals for over a month. > Besides, you can do this in userland already anyway: >
haha :) Yes, you can do many things in userland.
> <?php > class InvalidTypeException extends Exception {} > class UnknownTypeException extends Exception {} > > class Types { > const INTEGER = 1; > const FLOAT = 2; > const STRING = 3; > const OBJECT = 4; > const BOOLEAN = 5; > > private $val, $type; > > public function __construct($val, $type) { > $this->type = $type; > $this->setValue($val); > } > public function setValue($val) { > switch($this->type) { > case self::INTEGER: > if (!is_int($val)) { > throw new InvalidTypeException; > } > break; > > case self::FLOAT: > if (!is_float($val)) { > throw new InvalidTypeException; > } > break; > > case self::STRING: > if (!is_string($val)) { > throw new InvalidTypeException; > } > break; > > case self::OBJECT: > if (!is_object($val)) { > throw new InvalidTypeException; > } > break; > > case self::BOOLEAN: > if (!is_bool($val)) { > throw new InvalidTypeException; > } > break; > > default: > throw new UnknownTypeException; > > } > $this->val = $val; > } > public function getValue() { > return $this->val; > } > public function __toString() { > return (string)$this->getValue(); > } > } > > class Integer extends Types { > public function __construct($val) { > parent::__construct($val, Types::INTEGER); > } > } > class String extends Types { > public function __construct($val) { > parent::__construct($val, Types::STRING); > } > } > class Float extends Types { > public function __construct($val) { > parent::__construct($val, Types::FLOAT); > } > } > class Object extends Types { > public function __construct($val) { > parent::__construct($val, Types::OBJECT); > } > } > class Boolean extends Types { > public function __construct($val) { > parent::__construct($val, Types::BOOLEAN); > } > } > function type_hint_integer(Integer $val) { > echo $val, "\n"; > } > function type_hint_string(String $val) { > echo $val, "\n"; > } > function type_hint_float(Float $val) { > echo $val, "\n"; > } > > > > type_hint_integer(new Integer(123)); > type_hint_string(new String("string")); > type_hint_float(new Float(0.25)); >
Nice implementation, so ? Still more code to include in your code. Anyways, the "Optional" part in the subject means ... optional. Which means that it does not *have* to be used and that the default hinting will still be loose. That gives the chance to anyone to use either strongly-typed code or loosely-typed code. You know as much as I do that it's quite easy to do (implement - like you just did), but if it's not in by default, people won't care about doing it or implementing it. On the other hand, if it's there, some people might be tempted to use it, and will. But hey.. php's a loosely typed language and it'll definitely stay that way. But only giving the choice to someone to use what he feels like is quite interesting in my opinion. Anyways, no need to get all grumpy, you can just say you don't like the idea, that'll do just as well. Anyways, that was a thought, it's quite easy to implement as an extension and would be light and simple for anyone to either use or ignore it.
> -Hannes >
-- David Coallier, Founder & Software Architect, Agora Production (http://agoraproduction.com) 51.42.06.70.18

Sam Barrow

18 years ago
I both like and dislike the fact that PHP is so loosely typed. It makes it very easy, which I like, however it is sometimes not strict enough, allowing for undetected errors, which of course I don't like. I think the ideal solution here is to make it a hybrid type of language, where typing is dynamic, but still controllable using type hinting. You can use whichever you want where you want. Functions can have 3 parameters, two hinted and one not. Keep in mind this is not strict typing. Key word: "hinting". If all is 100% optional, your application can take advantage of the flexibility of dynamic typing in places where it's needed, and also take advantage of the control that type hinting gives you where it's appropriate. Most languages either allow you to do whatever you want, making some code chaotic and some bugs harder to detect, whereas some are the complete opposite, imposing very strict limits on your variables, and severely limiting your flexibility. PHP could be the middle ground, with maximum flexibility, but also maximum control. I'm currently using my type hinting patch with PHP 5.3 for an application I'm working on with about 10,000 lines of code and hundreds of functions, big and small. I'm using type hinting in most of the functions, but not all. It's been very helpful so far, especially with the "scalar" and "number" type hints. On Sun, 2007-11-18 at 21:50 -0500, David Coallier wrote:

Richard Quadling

18 years ago
On 19/11/2007, Sam Barrow <sam@sambarrow.com> wrote:
> I both like and dislike the fact that PHP is so loosely typed. It makes > it very easy, which I like, however it is sometimes not strict enough, > allowing for undetected errors, which of course I don't like. > > I think the ideal solution here is to make it a hybrid type of language, > where typing is dynamic, but still controllable using type hinting. You > can use whichever you want where you want. Functions can have 3 > parameters, two hinted and one not. > > Keep in mind this is not strict typing. Key word: "hinting". > > If all is 100% optional, your application can take advantage of the > flexibility of dynamic typing in places where it's needed, and also take > advantage of the control that type hinting gives you where it's > appropriate. Most languages either allow you to do whatever you want, > making some code chaotic and some bugs harder to detect, whereas some > are the complete opposite, imposing very strict limits on your > variables, and severely limiting your flexibility. PHP could be the > middle ground, with maximum flexibility, but also maximum control. > > I'm currently using my type hinting patch with PHP 5.3 for an > application I'm working on with about 10,000 lines of code and hundreds > of functions, big and small. I'm using type hinting in most of the > functions, but not all. It's been very helpful so far, especially with > the "scalar" and "number" type hints. > > On Sun, 2007-11-18 at 21:50 -0500, David Coallier wrote: > > On Nov 18, 2007 7:24 PM, Hannes Magnusson <hannes.magnusson@gmail.com> wrote: > > > On Nov 19, 2007 12:13 AM, David Coallier <davidc@php.net> wrote: > > > > I was thinking at something along the lines of objects also for instance: > > > > > > > > $i = new Integer(33); > > > > > > > > function foo(Integer $var) { > > > > } > > > > > > > > foo ($i); else it emits a fatal error. But also if you do > > > > > > > > $i = "Name"; that would emit a fatal error because the value is > > > > suposed to be an int. This might look a bit too much like java, but as > > > > an extension it could be something quite interesting I believe. > > > > > > > > String, Object, Integer, Scalar, Float and what else. > > > > > > > > So thinking of something like > > > > > > > > $string = new String("Foo"); > > > > $string = "bar" or $string->setValue("Bar"); would do > > > > > > > > $float = new Float(4.242); > > > > $float->setValue('foobar'); // That emits an error > > > > $float->setValue(3.14159); > > > > > > > > echo $float; (__toString) or echo $float->getValue; to echo it's content/value > > > > > > > > and so on. > > > > > > That has got to be the worst idea I've heard on internals for over a month. > > > Besides, you can do this in userland already anyway: > > > > > > > haha :) Yes, you can do many things in userland. > > > > > <?php > > > class InvalidTypeException extends Exception {} > > > class UnknownTypeException extends Exception {} > > > > > > class Types { > > > const INTEGER = 1; > > > const FLOAT = 2; > > > const STRING = 3; > > > const OBJECT = 4; > > > const BOOLEAN = 5; > > > > > > private $val, $type; > > > > > > public function __construct($val, $type) { > > > $this->type = $type; > > > $this->setValue($val); > > > } > > > public function setValue($val) { > > > switch($this->type) { > > > case self::INTEGER: > > > if (!is_int($val)) { > > > throw new InvalidTypeException; > > > } > > > break; > > > > > > case self::FLOAT: > > > if (!is_float($val)) { > > > throw new InvalidTypeException; > > > } > > > break; > > > > > > case self::STRING: > > > if (!is_string($val)) { > > > throw new InvalidTypeException; > > > } > > > break; > > > > > > case self::OBJECT: > > > if (!is_object($val)) { > > > throw new InvalidTypeException; > > > } > > > break; > > > > > > case self::BOOLEAN: > > > if (!is_bool($val)) { > > > throw new InvalidTypeException; > > > } > > > break; > > > > > > default: > > > throw new UnknownTypeException; > > > > > > } > > > $this->val = $val; > > > } > > > public function getValue() { > > > return $this->val; > > > } > > > public function __toString() { > > > return (string)$this->getValue(); > > > } > > > } > > > > > > class Integer extends Types { > > > public function __construct($val) { > > > parent::__construct($val, Types::INTEGER); > > > } > > > } > > > class String extends Types { > > > public function __construct($val) { > > > parent::__construct($val, Types::STRING); > > > } > > > } > > > class Float extends Types { > > > public function __construct($val) { > > > parent::__construct($val, Types::FLOAT); > > > } > > > } > > > class Object extends Types { > > > public function __construct($val) { > > > parent::__construct($val, Types::OBJECT); > > > } > > > } > > > class Boolean extends Types { > > > public function __construct($val) { > > > parent::__construct($val, Types::BOOLEAN); > > > } > > > } > > > function type_hint_integer(Integer $val) { > > > echo $val, "\n"; > > > } > > > function type_hint_string(String $val) { > > > echo $val, "\n"; > > > } > > > function type_hint_float(Float $val) { > > > echo $val, "\n"; > > > } > > > > > > > > > > > > type_hint_integer(new Integer(123)); > > > type_hint_string(new String("string")); > > > type_hint_float(new Float(0.25)); > > > > > > > Nice implementation, so ? Still more code to include in your code. > > Anyways, the "Optional" part in the subject means ... optional. Which > > means that it does not *have* to be used and that the default hinting > > will still be loose. > > > > That gives the chance to anyone to use either strongly-typed code or > > loosely-typed code. You know as much as I do that it's quite easy to > > do (implement - like you just did), but if it's not in by default, > > people won't care about doing it or implementing it. > > > > On the other hand, if it's there, some people might be tempted to use > > it, and will. But hey.. php's a loosely typed language and it'll > > definitely stay that way. But only giving the choice to someone to use > > what he feels like is quite interesting in my opinion. > > > > Anyways, no need to get all grumpy, you can just say you don't like > > the idea, that'll do just as well. Anyways, that was a thought, it's > > quite easy to implement as an extension and would be light and simple > > for anyone to either use or ignore it. > > > > > > > -Hannes > > > > > > > > > > > -- > > David Coallier, > > Founder & Software Architect, > > Agora Production (http://agoraproduction.com) > > 51.42.06.70.18
What if type hinting just generated an E_NOTICE. Nothing more for the time being. Call it an experimental option. I already use hungarian notation for all my params ( cause I'm unimaginative with my var names I suppose!), so the type hinting would help me enforce this and reduce errors in the long run (code goes from me to other developers who only glance at the dox)
-- ----- Richard Quadling Zend Certified Engineer : http://zend.com/zce.php?c=ZEND002498&r=213474731 "Standing on the shoulders of some very clever giants!"

Alexey Zakhlestin

18 years ago
On 11/19/07, Richard Quadling <rquadling@googlemail.com> wrote:
> What if type hinting just generated an E_NOTICE. Nothing more for the > time being. > > Call it an experimental option. > > I already use hungarian notation for all my params ( cause I'm > unimaginative with my var names I suppose!), so the type hinting would > help me enforce this and reduce errors in the long run (code goes from > me to other developers who only glance at the dox)
I like this idea but E_STRICT seems to fit better :)
-- Alexey Zakhlestin http://blog.milkfarmsoft.com/

Sam Barrow

18 years ago
Good idea about reducing the error warning level, isn't this currently fatal? We could just turn it into an e_notice or e_warning. On Mon, 2007-11-19 at 11:37 +0000, Richard Quadling wrote:

Sam Barrow

18 years ago
What is the general opinion on multiple class inheritance. I have a need for it. I have objects for all user input fields. $username = new field ; $username -> name = 'username' ; $username -> maxLen = 32 ; I have three types of fields. Fields that are automatically put in the database, such as timestamps, fields that are inputted but not stored, such as "confirm password", and fields that are inputted by the user AND stored in the database, such as username and password. Now i have 3 classes: - abstractField (has methods and properties that apply to all fields). - inputField, extends abstractField (has methods and properties for display of input form elements and labels). - dbField, extends abstractField (has methods for storing and retrieving in db, etc.). However for fields that are inputted AND stored in the db, i need to extend both inputField and dbField. - inputDbField extends inputField, dbField. Sure, there may be quick hacks to do this, but the only proper way seems to be to extend both classes, and I don't want to duplicate any code (dbField and inputField are both pretty big, and any modifications will also have to be replicated). And no, I don't want to use interfaces. Interfaces will barely do anything for me, I'll still have to duplicate my method bodies, and properties.

Edward Z. Yang

18 years ago
Sam Barrow wrote:
> What is the general opinion on multiple class inheritance. I have a need > for it.
Use object composition?
-- Edward Z. Yang GnuPG: 0x869C48DA HTML Purifier <http://htmlpurifier.org> Anti-XSS Filter [[ 3FA8 E9A9 7385 B691 A6FC B3CB A933 BE7D 869C 48DA ]]

Larry Garfield

18 years ago
It sounds like you want to be using decorators instead. http://en.wikipedia.org/wiki/Decorator_pattern On Sunday 18 November 2007, Sam Barrow wrote:
> What is the general opinion on multiple class inheritance. I have a need > for it. I have objects for all user input fields. > > $username = new field ; > $username -> name = 'username' ; > $username -> maxLen = 32 ; > > I have three types of fields. Fields that are automatically put in the > database, such as timestamps, fields that are inputted but not stored, > such as "confirm password", and fields that are inputted by the user AND > stored in the database, such as username and password. > > Now i have 3 classes: > > - abstractField (has methods and properties that apply to all fields). > - inputField, extends abstractField (has methods and properties for > display of input form elements and labels). > - dbField, extends abstractField (has methods for storing and retrieving > in db, etc.). > > However for fields that are inputted AND stored in the db, i need to > extend both inputField and dbField. > > - inputDbField extends inputField, dbField. > > Sure, there may be quick hacks to do this, but the only proper way seems > to be to extend both classes, and I don't want to duplicate any code > (dbField and inputField are both pretty big, and any modifications will > also have to be replicated). > > And no, I don't want to use interfaces. Interfaces will barely do > anything for me, I'll still have to duplicate my method bodies, and > properties.
-- Larry Garfield AIM: LOLG42 larry@garfieldtech.com ICQ: 6817012 "If nature has made any one thing less susceptible than all others of exclusive property, it is the action of the thinking power called an idea, which an individual may exclusively possess as long as he keeps it to himself; but the moment it is divulged, it forces itself into the possession of every one, and the receiver cannot dispossess himself of it." -- Thomas Jefferson

Edward Z. Yang

18 years ago
Larry Garfield wrote:
> It sounds like you want to be using decorators instead.
The decorator pattern is inappropriate for this case, because Sam wants to extend the interface, not change the behavior of an existing one.
-- Edward Z. Yang GnuPG: 0x869C48DA HTML Purifier <http://htmlpurifier.org> Anti-XSS Filter [[ 3FA8 E9A9 7385 B691 A6FC B3CB A933 BE7D 869C 48DA ]]

Larry Garfield

18 years ago
(Sorry, hit reply too soon.) Or, alternatively, you can mostly implement "friend" functions of a sort: http://www.garfieldtech.com/blog/php-magic-call but they have a performance penalty: http://www.garfieldtech.com/blog/magic-benchmarks On Sunday 18 November 2007, Sam Barrow wrote:
> What is the general opinion on multiple class inheritance. I have a need > for it. I have objects for all user input fields. > > $username = new field ; > $username -> name = 'username' ; > $username -> maxLen = 32 ; > > I have three types of fields. Fields that are automatically put in the > database, such as timestamps, fields that are inputted but not stored, > such as "confirm password", and fields that are inputted by the user AND > stored in the database, such as username and password. > > Now i have 3 classes: > > - abstractField (has methods and properties that apply to all fields). > - inputField, extends abstractField (has methods and properties for > display of input form elements and labels). > - dbField, extends abstractField (has methods for storing and retrieving > in db, etc.). > > However for fields that are inputted AND stored in the db, i need to > extend both inputField and dbField. > > - inputDbField extends inputField, dbField. > > Sure, there may be quick hacks to do this, but the only proper way seems > to be to extend both classes, and I don't want to duplicate any code > (dbField and inputField are both pretty big, and any modifications will > also have to be replicated). > > And no, I don't want to use interfaces. Interfaces will barely do > anything for me, I'll still have to duplicate my method bodies, and > properties.
-- Larry Garfield AIM: LOLG42 larry@garfieldtech.com ICQ: 6817012 "If nature has made any one thing less susceptible than all others of exclusive property, it is the action of the thinking power called an idea, which an individual may exclusively possess as long as he keeps it to himself; but the moment it is divulged, it forces itself into the possession of every one, and the receiver cannot dispossess himself of it." -- Thomas Jefferson

Sam Barrow

18 years ago
I was considering something like this, just seems like more of a hack to me than the right way to do it, plus the performance hit you mentioned. On Mon, 2007-11-19 at 00:11 -0600, Larry Garfield wrote:

Sebastian Bergmann

18 years ago
Sam Barrow schrieb:
> What is the general opinion on multiple class inheritance.
It is a concept that only works correctly in CLOS?
-- Sebastian Bergmann http://sebastian-bergmann.de/ GnuPG Key: 0xB85B5D69 / 27A7 2B14 09E4 98CD 6277 0E5B 6867 C514 B85B 5D69

Sam Barrow

18 years ago
I've been working on your patch, adding types and making it more stable. Worked pretty good for me though, with the exception of a couple things (some code was different in 5.3 than in your diff files, but that makes sense as you wrote this patch in 2006). I think more type hinting in all areas would be good, and if this is something that is going to be implemented, I am willing to do as much of the work as I'd have to. On Sun, 2007-11-18 at 21:25 +0100, Derick Rethans wrote:

Baptiste Autin

18 years ago
I know well the sacred "loose typing" anthem, but is there any hope to see the class properties type hinted one day? (and the return value as well) I mean, as an OPTION, as it is already done with function parameters... Class A { } Class B { public A $a; } Because: 1. It is so strange to see that in PHP5 only the function parameters have this feature implemented today. It's like a half-done job. 2. Many design patterns make use of composition (versus inheritance - thanks the GoF) and a composition would be more readable with such a feature. 3. Model-driven, reverse-enginering tools would appreciate this too. Other question: would it be conceivable to gather the ENV / GET / POST (and so on) variables into one or more system classes? (sorts of "context classes") so that one would always have the choice between looking into the traditional arrays $_xxx, or into the system classes (I'm sure this point has already been discussed, but I wasn't there :-).

Sam Barrow

18 years ago
I think type hinting of all variables eventually would be great, as long as it stays completely optional. On Mon, 2007-11-19 at 01:18 +0100, Baptiste Autin wrote:

Cristian Rodriguez

18 years ago
2007/11/15, Sam Barrow <sam@sambarrow.com>:
> > I found a patch by Derick online to allow for scalar type hinting, and > made it work with the newest snapshot of PHP 5.3.
IIRC Hannes had a patch to implement this the right way, but unfortunately it has not been merged. . Hopefully he can publish an updated version somewhere because I agree that having this functionality is a good thing.

David Coallier

18 years ago
We actually had spoken of that on irc a few times and a while ago, the answer I got back was basically, "If you want java, use java"... On Nov 16, 2007 8:40 PM, Cristian Rodriguez <judas.iscariote@gmail.com> wrote:
> 2007/11/15, Sam Barrow <sam@sambarrow.com>: > > > > I found a patch by Derick online to allow for scalar type hinting, and > > made it work with the newest snapshot of PHP 5.3. > > IIRC Hannes had a patch to implement this the right way, but > unfortunately it has not been merged. > . > Hopefully he can publish an updated version somewhere because I agree > that having this functionality is a good thing. > > > -- > PHP Internals - PHP Runtime Development Mailing List > To unsubscribe, visit: http://www.php.net/unsub.php > >
-- David Coallier, Founder & Software Architect, Agora Production (http://agoraproduction.com) 51.42.06.70.18

Marcus Börger

18 years ago
Hello Sam, since the thread turned to autoboxing. You guys may want to have a look at Pecl/SPL_Types which provides the base functionality to implement autoboxing already. Right now it provides only a Bool and an Enum class though. marcus Thursday, November 15, 2007, 4:27:17 PM, you wrote:
> I found a patch by Derick online to allow for scalar type hinting, and > made it work with the newest snapshot of PHP 5.3. I also added the > ability to type hint for resources. I would like to ask that it be added > to the next PHP release. It allows type hinting for int, float, bool, > string, resource, and object, I also added the ability to use the > secondary keywords for all of these types (such as double, real, long, > etc.).
> It will maintain 100% backwards compatibility, as the type hinting is > 100% optional, implemented in the same way as array/class type hinting.
> I have the patch on my PC, please let me know where and when i can > submit it. I'd be happy to do the patching and submission myself, just > asking for permission here.
Best regards, Marcus