YANP (Yet Another Namespace Proposal) for PHP

[MiiJaySung]

You asked

Actually, it is a real question. I wish Ruby could make a cup of coffee. The VM isn't there yet, still too slow and too big for embedding in a coffee machine. I think someone did get it running on a PDA though, which is heading in the right direction.

Douglas

If ya want coffee then Sun Java has lots of pretty coffee mugs :P Sure enough, you could even use Java for real coffee machine. Of course it would need to be powered by a Quad processor SPARC server. Which I guess ain't too bad if you're a multi billion dollar corporation with shed loads of cash to throw away.

Returning back to namespaces, I agree with Marcus, I don't understand the point of the private class cack, it's much like the final keyword, it's totally evil and no one has any use for it in PHP anyway

[BerislavLopac]

Sure enough, you could even use Java for real coffee machine. Of course it would need to be powered by a Quad processor SPARC server.

Had no idea that my mobile phone is powered by a Quad processor SPARC server...

[bonefry]

If ya want coffee then Sun Java has lots of pretty coffee mugs :P Sure enough, you could even use Java for real coffee machine. Of course it would need to be powered by a Quad processor SPARC server. Which I guess ain't too bad if you're a multi billion dollar corporation with shed loads of cash to throw away.

I see you developed some kind of Ruby sindrome were you bash Java with every opportunity. The problem is you ussually don't know what you are talking about (joke or not). And I have to defend it because I love it.

For raw performance take a look here: http://www.bytonic.de/html/benchmarks.html
I couldn't find a recent benchmark, but a good example is always http://www.walmart.com/ which handles according to them 7 million sessions and 106 million page views per day and it is based on Tomcat and Apache (no fancy websphere or weblogic or jboss).

Next time please provide some real arguments to support your claim.

Ruby doesn't even compile to bytecode (at least Python does, and PHP does too with some help). So you can't expect features like just in time compiling or a HotSpot compiler pretty soon.

[Dr Livingston]

Had no idea that my mobile phone is powered by a Quad processor SPARC server...

+99 to BerislavLopac. Cool...

[pachanga]

For a start, why does everything need strict static typing in a bytecode / "scripting"

Sometimes i find static typing very useful when reading someone else's code.
I'm currently studying Hibernate code(i think it's a glory of OOP) and this Java verbosity helps to understand its internals much better.

I really miss static typing in PHP, especially when i need to make some major refactoring(e.g changing method signature, etc) which spans many files. All modern editors simplify some tedious refactoring analyzing the code and using static types as helpers, i wonder if PHP type hinting would allow that.

[wdeboer]

Well, weak typing is just bad, I hope they will throw that support away in PHP 5.1 or make weak typing only available through a flag, which is off by default.

[Ryan Wray]

Nevermind, I see a limitation in what I posted.

[BerislavLopac]

Well, weak typing is just bad, I hope they will throw that support away in PHP 5.1 or make weak typing only available through a flag, which is off by default.

PHP has strong typing, not weak typing. On the other hand, it has dynamic typing as opposed to static typing.

[bonefry]

PHP has strong typing, not weak typing. On the other hand, it has dynamic typing as opposed to static typing.

What ? Strong Typing ?
That's sweet !

a = 12 + "24"
it's not strong typing
also

a = b + c
where b and c haven't been initialized, that's not strong typing either.

[BerislavLopac]

What ? Strong Typing ?
That's sweet !

a = 12 + "24"
it's not strong typing
also

a = b + c
where b and c haven't been initialized, that's not strong typing either.

That's dynamic typing, not weak typing. Look http://en.wikipedia.org/wiki/Strong_typing -- especially point 1.

[Ryan Wray]

What ? Strong Typing ?
That's sweet !

a = 12 + "24"
it's not strong typing
also

a = b + c
where b and c haven't been initialized, that's not strong typing either.

In the first example, it still has the types, it is just dynamic (ie. automatic type casting). Hence, what BerislavLopac said can apply to that statement (although, I would agree with you and say that PHP is not a strongly-typed language). As for the second example, well, I agree.

[bonefry]

That's dynamic typing, not weak typing. Look http://en.wikipedia.org/wiki/Strong_typing -- especially point 1.

No dude, you got it ALL wrong.

In a dynamic typed environment, messages sent to objects are not checked on compile-time, but rather on runtime.

And yes, your example is very good, and proves my point. Read it carefully one more time.

[bonefry]

In the first example, it still has the types, it is just dynamic (ie. automatic type casting). Hence, what BerislavLopac said can apply to that statement (although, I would agree with you and say that PHP is a strongly-typed language). As for the second example, well, I agree.

The first example is actually the most important one. Quote from the Wikipedia link BerislavLopac gave:

"A language is strongly typed if conversions between different types are forbidden. If such conversions are allowed, it is weakly typed."

[BerislavLopac]

However, the most important quote on that page is definitely:

Programming language expert Benjamin C. Pierce has said:

I spent a few weeks . . . trying to sort out the terminology of "strongly typed," "statically typed," "safe," etc., and found it amazingly difficult. . . . The usage of these terms is so various as to render them almost useless.

bonefry, I admit that I have fallen into the same trap you seem to and believed that the way I use those terms are the en-all and be-all definitions. But, like so many things in software industry, same things mean different things to different people. I have accepted a certain terminology and yes, it would be nice if everyone would agree with me, but I have to accept the fact that they don't.

In my usage:

• Strongly typed (or simply "typed"): when a language has types internally (examples: PHP, Python, Java, C).
• Weakly typed (or "untyped"): when a language stores the values without caring about types (e.g. all values are strings); each operation must internally resolve the implicit type to work with a variable (examples: TCL, Net.Data).
• Statically typed: when a type is bound to the variable as well as to its value. (examples: Java, C).
• Dynamically typed: when a type is bound to values only, and the actual type required for an operation is determined and the value is dynamically converted to it (examples: PHP, Python).

[bonefry]

However, the most important quote on that page is definitely:

Yes, but that does not mean they don't matter.
Because in Python 12 + "24" isn't allowed. Also in Python, you can't use a variable if it wasn't initialized.
And PHP is definitelly weak typed from all points if view if such term exists.

[DougBTX]

PHP has strong typing, not weak typing. On the other hand, it has dynamic typing as opposed to static typing.

PHP has loose and dynamic typing.

Loose:
$seven = 3 + '4';

Strong:
seven = 3 + '4'.to_i

Static:
seven := "seven";
seven := 7; // exception

Dynamic:
$seven = 'seven';
$seven = 7; // no exception

PHP, Ruby, Delphi, PHP respectively.

Later,
Douglas

[BerislavLopac]

Each of us is correct. And each of us is incorrect. Nobody is the final authority. Pot-a-to, potato.

[BerislavLopac]

So let's return this discussion to the topic of PHP namespaces. Would it really be so difficult to make an extension with namespaces as proposed by Marcus?

[DougBTX]

So let's return this discussion to the topic of PHP namespaces. Would it really be so difficult to make an extension with namespaces as proposed by Marcus?

Hmm, as Marcus says, "everything is done at parsing time". So, a proof of concept could just be a function like this:

function require_in($file, $namespace) {
  $content = file_get_contents($file);
  $content = str_replace('all_class_names',
                         "$namespace_all_class_names",
                         $content);
  eval($content);
} 

The tricky line is line 3, any ideas?

Once we have that, we can worry about adding the _once_ in require_once_in, then we can worry about caching the whole lot to the file systm so we can do require $file if file_exists.

Douglas

[nacho]

Wouldn't be easier to use a Register/ObjectFactory and store the registered objects under a common array key that would function as a fake namespace?

[Ryan Wray]

Hmm, as Marcus says, "everything is done at parsing time". So, a proof of concept could just be a function like this:

function require_in($file, $namespace) {
  $content = file_get_contents($file);
  $content = str_replace('all_class_names',
                         "$namespace_all_class_names",
                         $content);
  eval($content);
} 

The tricky line is line 3, any ideas?

Once we have that, we can worry about adding the _once_ in require_once_in, then we can worry about caching the whole lot to the file systm so we can do require $file if file_exists.

Douglas

I don't imagine the overhead of doing like that would be worth it. But, if that was the way, maybe regular expressions could probably be used to work it all out.

preg_replace('/^class (.*)/', 'class '. $ns .'_$1', $content); 

Not a powerful enough expression (and probably not even proper syntax) , but you get what I mean.

Too much of a hack for my liking.

[DougBTX]

Wouldn't be easier to use a Register/ObjectFactory and store the registered objects under a common array key that would function as a fake namespace?

No, because that isn't the problem. Example:

index.php

require('foo.php');
require('bar.php'); 

foo.php

class Klass {
  function hi() {
    echo 'hi';
  }
} 

bar.php

class Klass {
  function bye() {
    echo 'bye';
  }
} 

That will cause a fatal when you get to the second "class Klass" due to a "redeclared class". You need to get in before the require for it to work.

I wonder if anything in the Simple Test mock library could handle this?

Douglas

[DougBTX]

Not a powerful enough expression (and probably not even proper syntax) , but you get what I mean.

It would work if you have a single class in each file which doesn't reference any other classes. Say you have two classes which reference eachother, and you want to import them both under the same namespace, you'll have to update internal $foo = new Foo(); lines of code too. But can you assume that you can update all new Klass() lines?

A better fix would be to extend the extension so that this code works:

namespace ns {
  require('foo.php');
}
$foo = new ns:Foo(); 

I've not looked closely enough to see how that would work with the current code.

Douglas

[nacho]

No, because that isn't the problem.

You're absolutely right, my mistake. I was trying to avoid the ugly eval function ...

[Ezku]

A better fix would be to extend the extension so that this code works:

namespace ns {
  require('foo.php');
}
$foo = new ns:Foo(); 

I've not looked closely enough to see how that would work with the current code.

Somehow I understood it would - there's no point to the whole thing if it doesn't.