Overriding class constants

[Gordon]

I want to provide a set of static functions in a superclass that work
with class constants defined in a decendant of that class.
Unfortunately I've run into a snag with this idea. Example:

class SuperClass
{
const CNST = 'Super class';

public static function getCnst ()
{
return self::CNST;
}
}

class SubClass extends SuperClass
{
const CNST = 'Sub class';
}

echo (SubClass::getC nst ());

I'd want the above code tou output "Sub Class", but instead it outputs
"Super class".

It would sem that static functions can't work with overridden
constants in subclasses, or am I missing something?

[Jerry Stuckle]

Re: Overriding class constants

Gordon wrote:

I want to provide a set of static functions in a superclass that work
with class constants defined in a decendant of that class.
Unfortunately I've run into a snag with this idea. Example:
>
class SuperClass
{
const CNST = 'Super class';
>
public static function getCnst ()
{
return self::CNST;
}
}
>
class SubClass extends SuperClass
{
const CNST = 'Sub class';
}
>
echo (SubClass::getC nst ());
>
I'd want the above code tou output "Sub Class", but instead it outputs
"Super class".
>
It would sem that static functions can't work with overridden
constants in subclasses, or am I missing something?
>

Nope, unfortunately, this is true.

--
=============== ===
Remove the "x" from my email address
Jerry Stuckle
JDS Computer Training Corp.
jstucklex@attgl obal.net
=============== ===

[Michael Fesser]

Re: Overriding class constants

..oO(Gordon)

>I want to provide a set of static functions in a superclass that work
>with class constants defined in a decendant of that class.
>Unfortunatel y I've run into a snag with this idea. Example:
>
>class SuperClass
>{
> const CNST = 'Super class';
>
> public static function getCnst ()
> {
> return self::CNST;
> }
>}
>
>class SubClass extends SuperClass
>{
> const CNST = 'Sub class';
>}
>
>echo (SubClass::getC nst ());
>
>I'd want the above code tou output "Sub Class", but instead it outputs
>"Super class".
>
>It would sem that static functions can't work with overridden
>constants in subclasses, or am I missing something?

Currently static references are resolved at compile time. What you need
here is called late static binding and scheduled for PHP 5.3.

Micha

[Gordon]

Re: Overriding class constants

On Jul 3, 4:29 pm, Michael Fesser <neti...@gmx.de wrote:

.oO(Gordon)
>
>
>

I want to provide a set of static functions in a superclass that work
with class constants defined in a decendant of that class.
Unfortunately I've run into a snag with this idea. Example:

>

class SuperClass
{
const CNST = 'Super class';

>

public static function getCnst ()
{
return self::CNST;
}
}

>

class SubClass extends SuperClass
{
const CNST = 'Sub class';
}

>

echo (SubClass::getC nst ());

>

I'd want the above code tou output "Sub Class", but instead it outputs
"Super class".

>

It would sem that static functions can't work with overridden
constants in subclasses, or am I missing something?

>
Currently static references are resolved at compile time. What you need
here is called late static binding and scheduled for PHP 5.3.
>
Micha

Thanks.

Is there any kind of a workaround possible in older versions? The
only way I've found to get it to work as intended is to "inherit from
the clipboard" and cut and paste the methods into all the classes.
Maintainence nightmare.

[Mathieu Maes]

Re: Overriding class constants

On 4 jul, 10:14, Gordon <gordon.mc...@n tlworld.comwrot e:

On Jul 3, 4:29 pm, Michael Fesser <neti...@gmx.de wrote:
>
>
>

.oO(Gordon)

>

>I want to provide a set of static functions in a superclass that work
>with class constants defined in a decendant of that class.
>Unfortunatel y I've run into a snag with this idea.  Example:

>

>class SuperClass
>{
   const CNST      = 'Super class';

>

   public static function getCnst ()
   {
           return self::CNST;
   }
>}

>

>class SubClass extends SuperClass
>{
   const CNST      = 'Sub class';
>}

>

>echo (SubClass::getC nst ());

>

>I'd want the above code tou output "Sub Class", but instead it outputs
>"Super class".

>

>It would sem that static functions can't work with overridden
>constants in subclasses, or am I missing something?

>

Currently static references are resolved at compile time. What you need
here is called late static binding and scheduled for PHP 5.3.

>

Micha

>
Thanks.
>
Is there any kind of a workaround possible in older versions?  The
only way I've found to get it to work as intended is to "inherit from
the clipboard" and cut and paste the methods into all the classes.
Maintainence nightmare.

Theoratically speaking, a constant that needs to change during
execution is not a constant.
Could you explain the context a bit more ? I don't really see what
you're trying to do in your "simple" example.

[Gordon]

Re: Overriding class constants

On Jul 4, 9:50 am, Mathieu Maes <mathieu.m...@g mail.comwrote:

On 4 jul, 10:14, Gordon <gordon.mc...@n tlworld.comwrot e:
>
>
>

On Jul 3, 4:29 pm, Michael Fesser <neti...@gmx.de wrote:

>

.oO(Gordon)

>

I want to provide a set of static functions in a superclass that work
with class constants defined in a decendant of that class.
Unfortunately I've run into a snag with this idea. Example:

>

class SuperClass
{
const CNST = 'Super class';

>

public static function getCnst ()
{
return self::CNST;
}
}

>

class SubClass extends SuperClass
{
const CNST = 'Sub class';
}

>

echo (SubClass::getC nst ());

>

I'd want the above code tou output "Sub Class", but instead it outputs
"Super class".

>

It would sem that static functions can't work with overridden
constants in subclasses, or am I missing something?

>

Currently static references are resolved at compile time. What you need
here is called late static binding and scheduled for PHP 5.3.

>

Micha

>

Thanks.

>

Is there any kind of a workaround possible in older versions? The
only way I've found to get it to work as intended is to "inherit from
the clipboard" and cut and paste the methods into all the classes.
Maintainence nightmare.

>
Theoratically speaking, a constant that needs to change during
execution is not a constant.
Could you explain the context a bit more ? I don't really see what
you're trying to do in your "simple" example.

The values don't have to change during execution. They just have to
be different for different classes.

What I need is to be able to specify some parameters in a class that
describe how it is to be dealt with. For example for creating a new
item in the database I have a page that loads a HTML form for the user
to populate. However, each different type of item has a different
form. For example a document needs a field for content but a folder
doesn't as a folder is just a container for other items. I want to
specify in a class constant attributes for things such as that. For
example, a Folder class could have a NEW_FORM constant set to
'newfolder.php' whereas a Document class could have a NEW_FORM with a
value of 'newdoc.php'.

You have to be able to know what these values are before you can
instantise a class, however. Of course you can do a switch statement
based on the class of the item you intend to create, but that means
that, for example, I later want to add a new Blog class, then as well
as creating the new class I also have to remember to add Blog as an
option in the switch statement. If the Blog class itself can define
what makes it different from the other classes the system has to work
with then I can elimate the switch statement.

Of course there is only one switch statement in the above example but
in the real app there is more than one script that has to deal with
different types of object. If I used switch statements in all of
those cases then the result would have poor maintainability .

As was stated earlier by another poster the next version of PHP will
have a solution to this problem, namely late static binding. I guess
it's a common enough problem that they needed to add that feature to
the new release. Of course PHP 5.3 could be days off or several
months off and was hoping to find a workaround in the meantime.

[Jerry Stuckle]

Re: Overriding class constants

Gordon wrote:

On Jul 4, 9:50 am, Mathieu Maes <mathieu.m...@g mail.comwrote:

>On 4 jul, 10:14, Gordon <gordon.mc...@n tlworld.comwrot e:
>>
>>
>>

>>On Jul 3, 4:29 pm, Michael Fesser <neti...@gmx.de wrote:
>>>.oO(Gordon )
>>>>I want to provide a set of static functions in a superclass that work
>>>>with class constants defined in a decendant of that class.
>>>>Unfortunate ly I've run into a snag with this idea. Example:
>>>>class SuperClass
>>>>{
>>>> const CNST = 'Super class';
>>>> public static function getCnst ()
>>>> {
>>>> return self::CNST;
>>>> }
>>>>}
>>>>class SubClass extends SuperClass
>>>>{
>>>> const CNST = 'Sub class';
>>>>}
>>>>echo (SubClass::getC nst ());
>>>>I'd want the above code tou output "Sub Class", but instead it outputs
>>>>"Super class".
>>>>It would sem that static functions can't work with overridden
>>>>constants in subclasses, or am I missing something?
>>>Currently static references are resolved at compile time. What you need
>>>here is called late static binding and scheduled for PHP 5.3.
>>>Micha
>>Thanks.
>>Is there any kind of a workaround possible in older versions? The
>>only way I've found to get it to work as intended is to "inherit from
>>the clipboard" and cut and paste the methods into all the classes.
>>Maintainenc e nightmare.

>Theoraticall y speaking, a constant that needs to change during
>execution is not a constant.
>Could you explain the context a bit more ? I don't really see what
>you're trying to do in your "simple" example.

>
The values don't have to change during execution. They just have to
be different for different classes.
>
What I need is to be able to specify some parameters in a class that
describe how it is to be dealt with. For example for creating a new
item in the database I have a page that loads a HTML form for the user
to populate. However, each different type of item has a different
form. For example a document needs a field for content but a folder
doesn't as a folder is just a container for other items. I want to
specify in a class constant attributes for things such as that. For
example, a Folder class could have a NEW_FORM constant set to
'newfolder.php' whereas a Document class could have a NEW_FORM with a
value of 'newdoc.php'.
>
You have to be able to know what these values are before you can
instantise a class, however. Of course you can do a switch statement
based on the class of the item you intend to create, but that means
that, for example, I later want to add a new Blog class, then as well
as creating the new class I also have to remember to add Blog as an
option in the switch statement. If the Blog class itself can define
what makes it different from the other classes the system has to work
with then I can elimate the switch statement.
>
Of course there is only one switch statement in the above example but
in the real app there is more than one script that has to deal with
different types of object. If I used switch statements in all of
those cases then the result would have poor maintainability .
>
As was stated earlier by another poster the next version of PHP will
have a solution to this problem, namely late static binding. I guess
it's a common enough problem that they needed to add that feature to
the new release. Of course PHP 5.3 could be days off or several
months off and was hoping to find a workaround in the meantime.
>

But in your example, your inheritance hierarchy is incorrect. In folder
is not a "typeof" document. And no, physical storage media doesn't count.

Derived classes should always be more specific examples of the base class.

--
=============== ===
Remove the "x" from my email address
Jerry Stuckle
JDS Computer Training Corp.
jstucklex@attgl obal.net
=============== ===

[Chad Burrus]

Re: Overriding class constants

On Jul 4, 4:14 am, Gordon <gordon.mc...@n tlworld.comwrot e:

On Jul 3, 4:29 pm, Michael Fesser <neti...@gmx.de wrote:
>
>
>

.oO(Gordon)

>

>I want to provide a set of static functions in a superclass that work
>with class constants defined in a decendant of that class.
>Unfortunatel y I've run into a snag with this idea.  Example:

>

>class SuperClass
>{
   const CNST      = 'Super class';

>

   public static function getCnst ()
   {
           return self::CNST;
   }
>}

>

>class SubClass extends SuperClass
>{
   const CNST      = 'Sub class';
>}

>

>echo (SubClass::getC nst ());

>

>I'd want the above code tou output "Sub Class", but instead it outputs
>"Super class".

>

>It would sem that static functions can't work with overridden
>constants in subclasses, or am I missing something?

>

Currently static references are resolved at compile time. What you need
here is called late static binding and scheduled for PHP 5.3.

>

Micha

>
Thanks.
>
Is there any kind of a workaround possible in older versions?  The
only way I've found to get it to work as intended is to "inherit from
the clipboard" and cut and paste the methods into all the classes.
Maintainence nightmare.

One workaround that works with instances of a class (which
unfortunately does not work with static calls like in your example) is
to use the get_class function to grab the class name of the current
object, and then build a string you can pass to the constant function.

However, since you want static functions, the only way I know of to
work around the problem is to cheat: override the getConst function in
SubClass and pass in the class name to the SuperClass function, which
will give you something like this:

class SuperClass
{
const CNST = 'Super class';

public static function getCnst ($class = NULL)
{
if (!$class)
{
$class = get_class();
}
return constant($class . '::CNST');
}

}

class SubClass extends SuperClass
{
const CNST = 'Sub class';

public static function getCnst ()
{
return parent::getCnst (get_class());
}

}

echo SubClass::getCn st();
// outputs "Sub class"

You somewhat avoid the "inherit by clipboard" maintenance issue, but
you still have to override the getCnst method in all of your
subclasses, which kind of defeats the whole point.

HTH

[Gordon]

Re: Overriding class constants

On Jul 4, 4:19 pm, Jerry Stuckle <jstuck...@attg lobal.netwrote:

Gordon wrote:

On Jul 4, 9:50 am, Mathieu Maes <mathieu.m...@g mail.comwrote:

On 4 jul, 10:14, Gordon <gordon.mc...@n tlworld.comwrot e:

>

>On Jul 3, 4:29 pm, Michael Fesser <neti...@gmx.de wrote:
>>.oO(Gordon)
>>>I want to provide a set of static functions in a superclass that work
>>>with class constants defined in a decendant of that class.
>>>Unfortunatel y I've run into a snag with this idea.  Example:
>>>class SuperClass
>>>{
>>>   const CNST      = 'Super class';
>>>   public static function getCnst ()
>>>   {
>>>           return self::CNST;
>>>   }
>>>}
>>>class SubClass extends SuperClass
>>>{
>>>   const CNST      = 'Sub class';
>>>}
>>>echo (SubClass::getC nst ());
>>>I'd want the above code tou output "Sub Class", but instead it outputs
>>>"Super class".
>>>It would sem that static functions can't work with overridden
>>>constants in subclasses, or am I missing something?
>>Currently static references are resolved at compile time. What you need
>>here is called late static binding and scheduled for PHP 5.3.
>>Micha
>Thanks.
>Is there any kind of a workaround possible in older versions?  The
>only way I've found to get it to work as intended is to "inherit from
>the clipboard" and cut and paste the methods into all the classes.
>Maintainence nightmare.
Theoratically speaking, a constant that needs to change during
execution is not a constant.
Could you explain the context a bit more ? I don't really see what
you're trying to do in your "simple" example.

>

The values don't have to change during execution.  They just have to
be different for different classes.

>

What I need is to be able to specify some parameters in a class that
describe how it is to be dealt with.  For example for creating a new
item in the database I have a page that loads a HTML form for the user
to populate.  However, each different type of item has a different
form.  For example a document needs a field for content but a folder
doesn't as a folder is just a container for other items.  I want to
specify in a class constant attributes for things such as that.  For
example, a Folder class could have a NEW_FORM constant set to
'newfolder.php' whereas a Document class could have a NEW_FORM with a
value of 'newdoc.php'.

>

You have to be able to know what these values are before you can
instantise a class, however.  Of course you can do a switch statement
based on the class of the item you intend to create, but that means
that, for example, I later want to add a new Blog class, then as well
as creating the new class I also have to remember to add Blog as an
option in the switch statement.  If the Blog class itself can define
what makes it different from the other classes the system has to work
with then I can elimate the switch statement.

>

Of course there is only one switch statement in the above example but
in the real app there is more than one script that has to deal with
different types of object.  If I used switch statements in all of
those cases then the result would have poor maintainability .

>

As was stated earlier by another poster the next version of PHP will
have a solution to this problem, namely late static binding.  I guess
it's a common enough problem that they needed to add that feature to
the new release.  Of course PHP 5.3 could be days off or several
months off and was hoping to find a workaround in the meantime.

>
But in your example, your inheritance hierarchy is incorrect.  In folder
is not a "typeof" document.  And no, physical storage media doesn't count.
>
Derived classes should always be more specific examples of the base class..
>
--
=============== ===
Remove the "x" from my email address
Jerry Stuckle
JDS Computer Training Corp.
jstuck...@attgl obal.net
=============== ===

Both Folder and Document inherit from a common ancestor, Item.
Everything in the system have some commonality, they all have a
numeric ID and numeric parent (though it can be 0 for root items, aka
Sites), they all have titles and filesystem names, and they all have
metadata for searching. There are other classes that also descend
from Item, called Template, and Asset. Folder has a descendant, Site
(A Site is a Folder that is always a root item, and when they are
created they need to have some system Folders set up). The next
edition will be a Survey class, that descends from Document. A Survey
behaves like a Document (it's basically a Document that contains a web
form in terms of presentation) but it also has to handle data
acquisition.

The idea is to define static methods in Item that work on the
constants defined in the various descendant classes. Of course until
late static binding is available that's not going to work.

[Gordon]

Re: Overriding class constants

On Jul 4, 7:01 pm, Chad Burrus <chad.bur...@la t-inc.comwrote:

On Jul 4, 4:14 am, Gordon <gordon.mc...@n tlworld.comwrot e:
>
>
>

On Jul 3, 4:29 pm, Michael Fesser <neti...@gmx.de wrote:

>

.oO(Gordon)

>

I want to provide a set of static functions in a superclass that work
with class constants defined in a decendant of that class.
Unfortunately I've run into a snag with this idea.  Example:

>

class SuperClass
{
   const CNST      = 'Super class';

>

   public static function getCnst ()
   {
           return self::CNST;
   }
}

>

class SubClass extends SuperClass
{
   const CNST      = 'Sub class';
}

>

echo (SubClass::getC nst ());

>

I'd want the above code tou output "Sub Class", but instead it outputs
"Super class".

>

It would sem that static functions can't work with overridden
constants in subclasses, or am I missing something?

>

Currently static references are resolved at compile time. What you need
here is called late static binding and scheduled for PHP 5.3.

>

Micha

>

Thanks.

>

Is there any kind of a workaround possible in older versions?  The
only way I've found to get it to work as intended is to "inherit from
the clipboard" and cut and paste the methods into all the classes.
Maintainence nightmare.

>
One workaround that works with instances of a class (which
unfortunately does not work with static calls like in your example) is
to use the get_class function to grab the class name of the current
object, and then build a string you can pass to the constant function.
>
However, since you want static functions, the only way I know of to
work around the problem is to cheat: override the getConst function in
SubClass and pass in the class name to the SuperClass function, which
will give you something like this:
>
class SuperClass
{
        const CNST = 'Super class';
>
        public static function getCnst ($class = NULL)
        {
                if (!$class)
                {
                        $class = get_class();
                }
                return constant($class . '::CNST');
        }
>
}
>
class SubClass extends SuperClass
{
        const CNST = 'Sub class';
>
        public static function getCnst ()
        {
                return parent::getCnst (get_class());
        }
>
}
>
echo SubClass::getCn st();
// outputs "Sub class"
>
You somewhat avoid the "inherit by clipboard" maintenance issue, but
you still have to override the getCnst method in all of your
subclasses, which kind of defeats the whole point.
>
HTH

Thanks, that's not a bad idea. While the functions still have to be in
all the classes only the ancestor class has to define the actual
logic, all the descendants only have to call it. That's certainly an
improvement.

[Jerry Stuckle]

Re: Overriding class constants

Gordon wrote:

On Jul 4, 4:19 pm, Jerry Stuckle <jstuck...@attg lobal.netwrote:

>Gordon wrote:

>>On Jul 4, 9:50 am, Mathieu Maes <mathieu.m...@g mail.comwrote:
>>>On 4 jul, 10:14, Gordon <gordon.mc...@n tlworld.comwrot e:
>>>>On Jul 3, 4:29 pm, Michael Fesser <neti...@gmx.de wrote:
>>>>>.oO(Gordon )
>>>>>>I want to provide a set of static functions in a superclass that work
>>>>>>with class constants defined in a decendant of that class.
>>>>>>Unfortuna tely I've run into a snag with this idea. Example:
>>>>>>class SuperClass
>>>>>>{
>>>>>> const CNST = 'Super class';
>>>>>> public static function getCnst ()
>>>>>> {
>>>>>> return self::CNST;
>>>>>> }
>>>>>>}
>>>>>>class SubClass extends SuperClass
>>>>>>{
>>>>>> const CNST = 'Sub class';
>>>>>>}
>>>>>>echo (SubClass::getC nst ());
>>>>>>I'd want the above code tou output "Sub Class", but instead it outputs
>>>>>>"Super class".
>>>>>>It would sem that static functions can't work with overridden
>>>>>>constan ts in subclasses, or am I missing something?
>>>>>Currentl y static references are resolved at compile time. What you need
>>>>>here is called late static binding and scheduled for PHP 5.3.
>>>>>Micha
>>>>Thanks.
>>>>Is there any kind of a workaround possible in older versions? The
>>>>only way I've found to get it to work as intended is to "inherit from
>>>>the clipboard" and cut and paste the methods into all the classes.
>>>>Maintainenc e nightmare.
>>>Theoraticall y speaking, a constant that needs to change during
>>>execution is not a constant.
>>>Could you explain the context a bit more ? I don't really see what
>>>you're trying to do in your "simple" example.
>>The values don't have to change during execution. They just have to
>>be different for different classes.
>>What I need is to be able to specify some parameters in a class that
>>describe how it is to be dealt with. For example for creating a new
>>item in the database I have a page that loads a HTML form for the user
>>to populate. However, each different type of item has a different
>>form. For example a document needs a field for content but a folder
>>doesn't as a folder is just a container for other items. I want to
>>specify in a class constant attributes for things such as that. For
>>example, a Folder class could have a NEW_FORM constant set to
>>'newfolder.ph p' whereas a Document class could have a NEW_FORM with a
>>value of 'newdoc.php'.
>>You have to be able to know what these values are before you can
>>instantise a class, however. Of course you can do a switch statement
>>based on the class of the item you intend to create, but that means
>>that, for example, I later want to add a new Blog class, then as well
>>as creating the new class I also have to remember to add Blog as an
>>option in the switch statement. If the Blog class itself can define
>>what makes it different from the other classes the system has to work
>>with then I can elimate the switch statement.
>>Of course there is only one switch statement in the above example but
>>in the real app there is more than one script that has to deal with
>>different types of object. If I used switch statements in all of
>>those cases then the result would have poor maintainability .
>>As was stated earlier by another poster the next version of PHP will
>>have a solution to this problem, namely late static binding. I guess
>>it's a common enough problem that they needed to add that feature to
>>the new release. Of course PHP 5.3 could be days off or several
>>months off and was hoping to find a workaround in the meantime.

>But in your example, your inheritance hierarchy is incorrect. In folder
>is not a "typeof" document. And no, physical storage media doesn't count.
>>
>Derived classes should always be more specific examples of the base class.
>>
>--
>============== ====
>Remove the "x" from my email address
>Jerry Stuckle
>JDS Computer Training Corp.
>jstuck...@attg lobal.net
>============== ====

>
Both Folder and Document inherit from a common ancestor, Item.
Everything in the system have some commonality, they all have a
numeric ID and numeric parent (though it can be 0 for root items, aka
Sites), they all have titles and filesystem names, and they all have
metadata for searching. There are other classes that also descend
from Item, called Template, and Asset. Folder has a descendant, Site
(A Site is a Folder that is always a root item, and when they are
created they need to have some system Folders set up). The next
edition will be a Survey class, that descends from Document. A Survey
behaves like a Document (it's basically a Document that contains a web
form in terms of presentation) but it also has to handle data
acquisition.
>
The idea is to define static methods in Item that work on the
constants defined in the various descendant classes. Of course until
late static binding is available that's not going to work.
>

It sounds like you're trying to derive everything from Item - which
really isn't a good idea. It ties all of your classes more closely
togehter, making them more dependent on each other.

Also, a Site is not a "type of" folder, although it can refer to a
folder Neither is a Survey a type of a Document. The results generated
from the survey, however, could be. Again, the document would be
related to the survey.

All in all, I think your hierarchy is going to cause you a lot more
problems than you see now.

--
=============== ===
Remove the "x" from my email address
Jerry Stuckle
JDS Computer Training Corp.
jstucklex@attgl obal.net
=============== ===

[Gordon]

Re: Overriding class constants

On Jul 5, 1:55 pm, Jerry Stuckle <jstuck...@attg lobal.netwrote:

Gordon wrote:

On Jul 4, 4:19 pm, Jerry Stuckle <jstuck...@attg lobal.netwrote:

Gordon wrote:
>On Jul 4, 9:50 am, Mathieu Maes <mathieu.m...@g mail.comwrote:
>>On 4 jul, 10:14, Gordon <gordon.mc...@n tlworld.comwrot e:
>>>On Jul 3, 4:29 pm, Michael Fesser <neti...@gmx.de wrote:
>>>>.oO(Gordo n)
>>>>>I want to provide a set of static functions in a superclass that work
>>>>>with class constants defined in a decendant of that class.
>>>>>Unfortunat ely I've run into a snag with this idea.  Example:
>>>>>class SuperClass
>>>>>{
>>>>>   const CNST      = 'Super class';
>>>>>   public static function getCnst ()
>>>>>   {
>>>>>           return self::CNST;
>>>>>   }
>>>>>}
>>>>>class SubClass extends SuperClass
>>>>>{
>>>>>   const CNST      = 'Sub class';
>>>>>}
>>>>>echo (SubClass::getC nst ());
>>>>>I'd want the above code tou output "Sub Class", but instead it outputs
>>>>>"Super class".
>>>>>It would sem that static functions can't work with overridden
>>>>>constant s in subclasses, or am I missing something?
>>>>Currently static references are resolved at compile time. What youneed
>>>>here is called late static binding and scheduled for PHP 5.3.
>>>>Micha
>>>Thanks.
>>>Is there any kind of a workaround possible in older versions?  The
>>>only way I've found to get it to work as intended is to "inherit from
>>>the clipboard" and cut and paste the methods into all the classes.
>>>Maintainen ce nightmare.
>>Theoratical ly speaking, a constant that needs to change during
>>execution is not a constant.
>>Could you explain the context a bit more ? I don't really see what
>>you're trying to do in your "simple" example.
>The values don't have to change during execution.  They just have to
>be different for different classes.
>What I need is to be able to specify some parameters in a class that
>describe how it is to be dealt with.  For example for creating a new
>item in the database I have a page that loads a HTML form for the user
>to populate.  However, each different type of item has a different
>form.  For example a document needs a field for content but a folder
>doesn't as a folder is just a container for other items.  I want to
>specify in a class constant attributes for things such as that.  For
>example, a Folder class could have a NEW_FORM constant set to
>'newfolder.php ' whereas a Document class could have a NEW_FORM with a
>value of 'newdoc.php'.
>You have to be able to know what these values are before you can
>instantise a class, however.  Of course you can do a switch statement
>based on the class of the item you intend to create, but that means
>that, for example, I later want to add a new Blog class, then as well
>as creating the new class I also have to remember to add Blog as an
>option in the switch statement.  If the Blog class itself can define
>what makes it different from the other classes the system has to work
>with then I can elimate the switch statement.
>Of course there is only one switch statement in the above example but
>in the real app there is more than one script that has to deal with
>different types of object.  If I used switch statements in all of
>those cases then the result would have poor maintainability .
>As was stated earlier by another poster the next version of PHP will
>have a solution to this problem, namely late static binding.  I guess
>it's a common enough problem that they needed to add that feature to
>the new release.  Of course PHP 5.3 could be days off or several
>months off and was hoping to find a workaround in the meantime.
But in your example, your inheritance hierarchy is incorrect.  In folder
is not a "typeof" document.  And no, physical storage media doesn't count.

>

Derived classes should always be more specific examples of the base class.

>

--
=============== ===
Remove the "x" from my email address
Jerry Stuckle
JDS Computer Training Corp.
jstuck...@attgl obal.net
=============== ===

>

Both Folder and Document inherit from a common ancestor, Item.
Everything in the system have some commonality, they all have a
numeric ID and numeric parent (though it can be 0 for root items, aka
Sites), they all have titles and filesystem names, and they all have
metadata for searching.  There are other classes that also descend
from Item, called Template, and Asset.  Folder has a descendant, Site
(A Site is a Folder that is always a root item, and when they are
created they need to have some system Folders set up). The next
edition will be a Survey class, that descends from Document. A Survey
behaves like a Document (it's basically a Document that contains a web
form in terms of presentation) but it also has to handle data
acquisition.

>

The idea is to define static methods in Item that work on the
constants defined in the various descendant classes.  Of course until
late static binding is available that's not going to work.

>
It sounds like you're trying to derive everything from Item - which
really isn't a good idea.  It ties all of your classes more closely
togehter, making them more dependent on each other.
>
Also, a Site is not a "type of" folder, although it can refer to a
folder Neither is a Survey a type of a Document.  The results generated
from the survey, however, could be.  Again, the document would be
related to the survey.
>
All in all, I think your hierarchy is going to cause you a lot more
problems than you see now.
>
--
=============== ===
Remove the "x" from my email address
Jerry Stuckle
JDS Computer Training Corp.
jstuck...@attgl obal.net
=============== ===

Conceptually a Site is a type of Folder, because both are represented
in the filesystem by directories and are containers for other items.
The Site is essentially a root Folder. The reason for the distinction
is that a Site is expected to contain a set of Folders the system uses
for storing uploaded files, images, etc, so its CreateItem method
creates them as well. As for the Survey, the interface it presents to
the outside world is a HTML page with a form in it, in other words a
Document (which represents a HTML page) with content appropriate to
the task in hand of collecting data. The results collected by the
survey are a different matter as they can be viewed different ways, as
a table, bar chart, pie chart, CSV file etc, and thus aren't really
conceptually Documents.

As for the Item class it handles all the tasks common to everything
managed by the system. They all have metadata, they all abstract
things that exist on the filesystem, etc. The Item class is the
biggest class, but it's not considerably bigger than the classes that
descend from it.

And as has already been established, the problem I'm having is down to
PHP not implementing functionality that I need for it, and that said
missing functionality will be in a future version.

[Jerry Stuckle]

Re: Overriding class constants

Gordon wrote:

On Jul 5, 1:55 pm, Jerry Stuckle <jstuck...@attg lobal.netwrote:

>Gordon wrote:

>>On Jul 4, 4:19 pm, Jerry Stuckle <jstuck...@attg lobal.netwrote:
>>>Gordon wrote:
>>>>On Jul 4, 9:50 am, Mathieu Maes <mathieu.m...@g mail.comwrote:
>>>>>On 4 jul, 10:14, Gordon <gordon.mc...@n tlworld.comwrot e:
>>>>>>On Jul 3, 4:29 pm, Michael Fesser <neti...@gmx.de wrote:
>>>>>>>.oO(Gord on)
>>>>>>>>I want to provide a set of static functions in a superclass that work
>>>>>>>>with class constants defined in a decendant of that class.
>>>>>>>>Unfortu nately I've run into a snag with this idea. Example:
>>>>>>>>class SuperClass
>>>>>>>>{
>>>>>>>> const CNST = 'Super class';
>>>>>>>> public static function getCnst ()
>>>>>>>> {
>>>>>>>> return self::CNST;
>>>>>>>> }
>>>>>>>>}
>>>>>>>>class SubClass extends SuperClass
>>>>>>>>{
>>>>>>>> const CNST = 'Sub class';
>>>>>>>>}
>>>>>>>>echo (SubClass::getC nst ());
>>>>>>>>I'd want the above code tou output "Sub Class", but instead it outputs
>>>>>>>>"Supe r class".
>>>>>>>>It would sem that static functions can't work with overridden
>>>>>>>>constan ts in subclasses, or am I missing something?
>>>>>>>Currentl y static references are resolved at compile time. What you need
>>>>>>>here is called late static binding and scheduled for PHP 5.3.
>>>>>>>Micha
>>>>>>Thanks.
>>>>>>Is there any kind of a workaround possible in older versions? The
>>>>>>only way I've found to get it to work as intended is to "inherit from
>>>>>>the clipboard" and cut and paste the methods into all the classes.
>>>>>>Maintaine nce nightmare.
>>>>>Theoratica lly speaking, a constant that needs to change during
>>>>>executio n is not a constant.
>>>>>Could you explain the context a bit more ? I don't really see what
>>>>>you're trying to do in your "simple" example.
>>>>The values don't have to change during execution. They just have to
>>>>be different for different classes.
>>>>What I need is to be able to specify some parameters in a class that
>>>>describe how it is to be dealt with. For example for creating a new
>>>>item in the database I have a page that loads a HTML form for the user
>>>>to populate. However, each different type of item has a different
>>>>form. For example a document needs a field for content but a folder
>>>>doesn't as a folder is just a container for other items. I want to
>>>>specify in a class constant attributes for things such as that. For
>>>>example, a Folder class could have a NEW_FORM constant set to
>>>>'newfolder. php' whereas a Document class could have a NEW_FORM with a
>>>>value of 'newdoc.php'.
>>>>You have to be able to know what these values are before you can
>>>>instantis e a class, however. Of course you can do a switch statement
>>>>based on the class of the item you intend to create, but that means
>>>>that, for example, I later want to add a new Blog class, then as well
>>>>as creating the new class I also have to remember to add Blog as an
>>>>option in the switch statement. If the Blog class itself can define
>>>>what makes it different from the other classes the system has to work
>>>>with then I can elimate the switch statement.
>>>>Of course there is only one switch statement in the above example but
>>>>in the real app there is more than one script that has to deal with
>>>>different types of object. If I used switch statements in all of
>>>>those cases then the result would have poor maintainability .
>>>>As was stated earlier by another poster the next version of PHP will
>>>>have a solution to this problem, namely late static binding. I guess
>>>>it's a common enough problem that they needed to add that feature to
>>>>the new release. Of course PHP 5.3 could be days off or several
>>>>months off and was hoping to find a workaround in the meantime.
>>>But in your example, your inheritance hierarchy is incorrect. In folder
>>>is not a "typeof" document. And no, physical storage media doesn't count.
>>>Derived classes should always be more specific examples of the base class.
>>>--
>>>============ ======
>>>Remove the "x" from my email address
>>>Jerry Stuckle
>>>JDS Computer Training Corp.
>>>jstuck...@at tglobal.net
>>>============ ======
>>Both Folder and Document inherit from a common ancestor, Item.
>>Everything in the system have some commonality, they all have a
>>numeric ID and numeric parent (though it can be 0 for root items, aka
>>Sites), they all have titles and filesystem names, and they all have
>>metadata for searching. There are other classes that also descend
>>from Item, called Template, and Asset. Folder has a descendant, Site
>>(A Site is a Folder that is always a root item, and when they are
>>created they need to have some system Folders set up). The next
>>edition will be a Survey class, that descends from Document. A Survey
>>behaves like a Document (it's basically a Document that contains a web
>>form in terms of presentation) but it also has to handle data
>>acquisition .
>>The idea is to define static methods in Item that work on the
>>constants defined in the various descendant classes. Of course until
>>late static binding is available that's not going to work.

>It sounds like you're trying to derive everything from Item - which
>really isn't a good idea. It ties all of your classes more closely
>togehter, making them more dependent on each other.
>>
>Also, a Site is not a "type of" folder, although it can refer to a
>folder Neither is a Survey a type of a Document. The results generated
>from the survey, however, could be. Again, the document would be
>related to the survey.
>>
>All in all, I think your hierarchy is going to cause you a lot more
>problems than you see now.
>>
>--
>============== ====
>Remove the "x" from my email address
>Jerry Stuckle
>JDS Computer Training Corp.
>jstuck...@attg lobal.net
>============== ====

>
Conceptually a Site is a type of Folder, because both are represented
in the filesystem by directories and are containers for other items.
The Site is essentially a root Folder. The reason for the distinction
is that a Site is expected to contain a set of Folders the system uses
for storing uploaded files, images, etc, so its CreateItem method
creates them as well. As for the Survey, the interface it presents to
the outside world is a HTML page with a form in it, in other words a
Document (which represents a HTML page) with content appropriate to
the task in hand of collecting data. The results collected by the
survey are a different matter as they can be viewed different ways, as
a table, bar chart, pie chart, CSV file etc, and thus aren't really
conceptually Documents.
>

Not so. The file system is only the way the data is stored. What if
you had a site which was entirely database driven? Then you have no
"file system" - just a single generic file and a .htaccess file.

Don't mix the physical representation and the overall design. It's a
common problem.

In a true OO environment, the site will NOT be a type of folder -
conceptually they are two entirely different types of objects.

As for the Item class it handles all the tasks common to everything
managed by the system. They all have metadata, they all abstract
things that exist on the filesystem, etc. The Item class is the
biggest class, but it's not considerably bigger than the classes that
descend from it.
>

Again, not really valid. From your design, almost everything would be
derived from Item - which means EVERYTHING it tied together. When you
have that type of relationship, it's much harder to make changes.

And as has already been established, the problem I'm having is down to
PHP not implementing functionality that I need for it, and that said
missing functionality will be in a future version.
>

No, you think the problem is not implementing the functionality you
need. I think you have an invalid inheritance hierarchy. You're only
seeing the initial problem (which, BTW, will occur in virtually any OO
language). You will have others in the future.

--
=============== ===
Remove the "x" from my email address
Jerry Stuckle
JDS Computer Training Corp.
jstucklex@attgl obal.net
=============== ===