Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:50206
Mailing-List: contact internals-help@lists.php.net; run by ezmlm
Received-SPF: pass (pb1.pair.com: domain gmail.com designates 209.85.161.42 as permitted sender)
DomainKey-Status: bad
DomainKey-Signature: a=rsa-sha1; c=nofws;
        d=gmail.com; s=gamma;
        h=mime-version:date:message-id:subject:from:to:content-type
         :content-transfer-encoding;
        b=GmlryHQbegZYFIKz8JG0kTwtCndwQj0SWTq5IQu1tlMtz5vMXDtD8v+D3HHfmcRQXB
         kZnQbnlsqsHQiQ43kD4U1F4URDsda23i2DobK3NgYNBIOiF0BJlpcaTc3qGEbE95kqEh
         +PTXg6yAsMkBOaenyYdxHwTGu/r6mzZ+ImZDQ=
MIME-Version: 1.0
Date: Mon, 15 Nov 2010 15:07:34 -0200
Message-ID: <AANLkTik703Bky+ZtjdzBMY8tr97m34jDSSMyoB9opJKR@mail.gmail.com>
To: PHP internals <internals@lists.php.net>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: quoted-printable
Subject: PHP 5.4 - Meta attribute (aka. Annotations) support discussion
From: guilhermeblanco@gmail.com ("guilhermeblanco@gmail.com")

Hi folks,

I'll start a series of topics (in this thread) about meta attribute
(aka. Annotations) discussion.
So as soon as we agree on each topic I'll open another point to be discusse=
d.
Only when we reach some consensus I'll open another topic discussion.
I suggest to have a poll for each topic, so we can do some democratic
internals decision. I'd reduce the scope of voters to only people with
PHP karma.

The first topic to be defined is: Does PHP need meta attribute support?

Any language is built at the top of syntax, words, semantics and context.
We cannot define it as a language if we don't have these 4 parts well
fit together.
For a simple sentence like: I am all ears, we can easily identify these par=
ts.
But the real meaning of this sentence (context) is only achieved with
another part of the language: signifier-significate.
A signifier is an atomic form, material element, visible, sensitive.
A significate is an idea, semantic content, conceptual element, not
sensitive perceptive.

This important part is only reminded when we get back to semantics,
specifically when talking about idiomatic expressions, a metaphor in
our sample. The meaning will only work if we have the context, a known
information. Our sentence would mean that spoker person only have
ears, while the true meaning is that he/she is paying attention.
A computer language doesn't differ from a language by any means. We
have tokens (words), a syntax, semantic. But the lack of
contextualization brings a hole on language and how your application
should behave. This behavior can be achieved by many different ways;
the simplest way is known as meta information.

Considering you have a lack of idiomatic expressions knowledge of
someone that is reading. He'd think person only have ears, but if you
meta notify him that it is a metaphor, he'd try to find a different
meaning. Example: (Metaphor) I am all ears.

Now that we find how meta-information can benefit languages, it is
time to understand how meta-information can benefit a computer
language.

A good example is define a Service that validates Domain Objects.
Imagining a scenario where you have a User class that holds email and
password. Email and password fields must be validated, for example,
email as an email and password with at least 6 chars. A generic
validation service could not have the User validation hardcoded inside
the service, so the solution would be an external factor that explains
to service how to validate the DO. One of the possible implementations
would be a UserValidator class that contains validation rules for each
field. This external factor is a good sample of meta-information. But
how would a meta attribute would solve this situation?
A Service could simple retrieve the validation instructions by
Reflecting the class which instance is going to be validated.
Basically, to validate a Domain Object, it is simply required to add
validation rules on desired attributes. Example:

class User {
    <Email Meta Information>
    protected $email;

    <At least 6 chars Meta Information>
    protected $password;
}

*NOTE*: Please understand that this do not enter in any implementation
details. It is not time (yet) to talk baout docblock, new syntax, etc.
Scope now is simpler than that.

Now that is possible to see how meta attribute support could benefit a
language, I compiled a simple list of known projects and how can they
benefit of meta support:
- phpUnit Providing meta functionality for test cases, examples:
@dataProvider for test data iteration, @expectedException for catching
exceptions, etc.
- Doctrine For Object-Relational mapping, examples: @Entity,
@OneToOne, @Id, etc.
- Zend Framework Server classes Used to automate mappings for XML-RPC,
SOAP, etc.
- FLOW3 for dependency injection and validation
- Symfony2 for validation and routing rules
- Others: Validation, Functional Behavior injection (which could take
advantage of Traits), etc.


Poll will be opened for voting for 7 days (1 week) starting from now.
Poll will be closed next Monday at 5pm GMT.
Question: Does PHP need meta attribute support?

Happy voting!

--=20
Guilherme Blanco
Mobile: +55 (16) 9215-8480
MSN: guilhermeblanco@hotmail.com
S=C3=A3o Paulo - SP/Brazil