[RFC] Constant Scalar Expressions (with constants)

Hi Bob,

thanks for your work, it’s awesome. One thing about constants:

Take this example:

const FOO = 2;

class Klass
{
const FOO = 1;
const BAR = FOO;
}

Here FOO is ambiguous, as it could resolve to 2 or to 1. If I understood
it correctly, global constants are not yet implemented but I would suggest
to always force prefixing the constant to get rid of the ambiguity and
allow for future extension.

const FOO = 2;

class Klass
{
const FOO = 1;
const BAR = self::FOO;
}

Does that make sense?

The code snippets you posted are already supported (e.g. in PHP 5.5). FOO
will resolve to the globals constant, self::FOO to the class constant.
Presumably the extended support in static expressions will retain those
semantics :)

Nikita

2013/11/7 Lars Strojny lars@strojny.net

Hi Bob,

thanks for your work, it’s awesome. One thing about constants:

Take this example:

const FOO = 2;

class Klass
{
const FOO = 1;
const BAR = FOO;
}

Here FOO is ambiguous, as it could resolve to 2 or to 1. If I understood
it correctly, global constants are not yet implemented but I would suggest
to always force prefixing the constant to get rid of the ambiguity and
allow for future extension.

const FOO = 2;

class Klass
{
const FOO = 1;
const BAR = self::FOO;
}

Does that make sense?

Yes, also from the consistency point of view: I'd expected to refer class
constants via its class name (or self/static). Having to ways to refer them
is confusing.

cu,
Lars

Hi!

As you all know, Anthony Ferrara had withdrawn his RFC a few months ago,
which was without support for constants in these expressions:

const foo = 10;
const bar = 10 * foo;

The use of a constant in a declaration of a constant was not yet
possible with his patch.

That's what I changed.

Enjoy reading the RFC at:

https://wiki.php.net/rfc/const_scalar_exprs

Bob Weinand

Special thanks go to Dmitry who had helped me a lot with feedback and a
deep review of the patch.

--
github.com/KingCrunch

Hi Lars,

In what way is this ambiguous?

I just tested; FOO never equals self::FOO. FOO always is a global constant, while self::FOO (and Klass::FOO) is the only way to refer to the class constant.

const FOO = 2;

class Klass
{
const FOO = 1;
const BAR = FOO; // value two, global constant, and if FOO is not defined, Klass::Bar === "FOO".
}

Bob

Am 7.11.2013 um 13:31 schrieb Lars Strojny lars@strojny.net:

Hi Bob,

thanks for your work, it’s awesome. One thing about constants:

Take this example:

const FOO = 2;

class Klass
{
const FOO = 1;
const BAR = FOO;
}

Here FOO is ambiguous, as it could resolve to 2 or to 1. If I understood it correctly, global constants are not yet implemented but I would suggest to always force prefixing the constant to get rid of the ambiguity and allow for future extension.

const FOO = 2;

class Klass
{
const FOO = 1;
const BAR = self::FOO;
}

Does that make sense?

cu,
Lars

Hi!

As you all know, Anthony Ferrara had withdrawn his RFC a few months ago, which was without support for constants in these expressions:

const foo = 10;
const bar = 10 * foo;

The use of a constant in a declaration of a constant was not yet possible with his patch.

That's what I changed.

Enjoy reading the RFC at:

https://wiki.php.net/rfc/const_scalar_exprs

Bob Weinand

Special thanks go to Dmitry who had helped me a lot with feedback and a deep review of the patch.

Am 07.11.2013 13:01 schrieb "Bob Weinand" bobwei9@hotmail.com:

https://wiki.php.net/rfc/const_scalar_exprs

Wow, this looks really nice!

Will this work with class autoloading, too? i.e. will

class foo {
const X = bar::X;
}

trigger autoloading for class bar?

Once again, the code you posted is already supported, not really part of
this proposal. And yes, it will invoke the autoloader, see
http://codepad.viper-7.com/2Xz8ux :)

Nikita

Am 7.11.2013 um 14:34 schrieb Derick Rethans derick@php.net:

Hi!

As you all know, Anthony Ferrara had withdrawn his RFC a few months ago, which was without support for constants in these expressions:

const foo = 10;
const bar = 10 * foo;

The use of a constant in a declaration of a constant was not yet possible with his patch.

That's what I changed.

Enjoy reading the RFC at:

https://wiki.php.net/rfc/const_scalar_exprs

https://github.com/bwoebi/php-src/compare/const_scalar_exprs#diff-4d0b93094aafe8cf519b4f0a3322f471L590

-#define IS_CALLABLE 10
+#define IS_CONSTANT_AST 10
+#define IS_CALLABLE 11

Can I please suggest that you do not change existing constant values?
You're using a macro to check for the constant type anyway, so you can
do the complex checking for constant types in there.

Just comparing twice (> and <) should be faster than comparing three times (three times an ==).

Dmitry had suggested this to me. And I don't see the big disadvantage here as IS_CALLABLE is anyway just used for the typehint. Nothing there which relies on the fact that IS_CALLABLE equals to 10.

I'm also of the opinion that just " This can allow for writing far
easier to understand code, by allowing for far more expressive code." is
by far not enough of a justification for a syntax addition. Please
please try to be convincing by having good arguments, presenting
alternatives and cons.

Well, that's the justification. I can also list other justifications, but that'd be just other ways to express the same, so I just kept it simple.

Alternatives… don't really exist. (at least not in function signatures for example)

And cons? Can't think of any problem with it.

Bob

cheers,
Derick

--
http://derickrethans.nl | http://xdebug.org
Like Xdebug? Consider a donation: http://xdebug.org/donate.php
twitter: @derickr and @xdebug
Posted with an email client that doesn't mangle email: alpine