====== PHP RFC: Your Title Here ======
  * Version: 0.9
  * Date: 2013-12-31
  * Author: Yasuo Ohgaki, yohgaki@ohgaki.net
  * Status: Draft
  * First Published at: http://wiki.php.net/rfc/gmp_number


===== Introduction =====

GMP arithmetic with normal operator is added by 
https://wiki.php.net/rfc/operator_overloading_gmp

GMP numbers work like normal PHP number(int/float) with normal PHP arithmetic operators now.

This RFC proposes GMP number compatible with PHP number(int/float).

Current GMP implementation does not support GMP float/rational numbers. For the sake of completeness, GMP float support is needed at least. There are functions to be modified to treat GMP number as normal PHP numbers.


===== Proposal =====

==== Add GMP float support to GMP module ====
<del>Add GMP float support to GMP module.

  * Add all GMP float API wrapper to module.
  * GMP float API wrapper supports PHP int/float and GMP int/float. Conversion is done automatically.
  * Default precision: 256 (?)</del>

Extracted out to a separate RFC: [[rfc:gmp-floating-point|GMP Floating Point Support]]

==== Add math function compatibility =====
Add math function compatibility.

  * Down grade GMP number (int/float) to PHP number (int/float) automatically.

Some other RFC may implement, full GMP number support in math functions, but this is out of this RFC scope.

==== Add variable function compatibility =====
Add variable function compatibility.

  * empty() - evaluated GMP int 0 and GMP float 0.0 as TRUE, otherwise FALSE.
  * is_scalar() - evaluated GMP number as TRUE.
  * (anything else?)



===== References =====

== Current empty() and is_scalar() behavior ==

[yohgaki@dev php-src]$ ./php-bin -a
Interactive shell

  php > $g = gmp_init(123);
  php > var_dump($v);
  object(GMP)#3 (1) {
    ["num"]=>
    string(1) "0"
  }
  php > $v = $g * 0;
  
  php > var_dump($v, empty($v), is_scalar($v));
  object(GMP)#4 (1) {
    ["num"]=>
    string(3) "0"
  }
  bool(false)
  bool(false)


== Current math function behavior ==

GMP float is not supported.
  php > $f = 2.2;
  php > $gf = gmp_init($f);
  
  Warning: gmp_init(): Unable to convert variable to GMP - wrong type in php shell code on line 1

max()/min()
  php > $g = gmp_init(123);
  php > var_dump($v);
  object(GMP)#4 (1) {
    ["num"]=>
    string(1) "0"
  }
  php > $g = gmp_init(5);
  php > var_dump(max(1,2,$g));
  object(GMP)#1 (1) {
    ["num"]=>
    string(1) "5"
  }
  php > var_dump(min(1,2,$g));
  int(1)
  php > var_dump(min($g, 6, 7));
  object(GMP)#1 (1) {
    ["num"]=>
    string(1) "5"
  }


== Current Ruby and PHP behavior ==
  [yohgaki@dev php-src]$ irb
  2.0.0p247 :001 > 10 / 3
   => 3 
  2.0.0p247 :002 > 10.0 / 3
   => 3.3333333333333335 
  2.0.0p247 :003 > 

  [yohgaki@dev php-src]$ php -a
  Interactive shell
  
  php > $g = gmp_init('10');
  php > echo $g / 3;
  3
  
  php > echo 10 / 3;
  3.3333333333333

===== Backward Incompatible Changes =====


===== Proposed PHP Version(s) =====

PHP 5.6 or later

===== Open Issues =====

  * Ways to implement GMP object evaluation in empty(), is_scalar()
    * Treat GMP object as special object
    * Implement scalar object
    * Implement autoboxing like feature
  * How GMP number should behave (Like PHP or Ruby?)

===== Future Scope =====

Math functions is better to support GMP float.

===== Proposed Voting Choices =====

Yes/No

===== Patches and Tests =====

I really appreciate if anyone could make patch for this RFC.

===== Implementation =====

After the project is implemented, this section should contain 
  - the version(s) it was merged to
  - a link to the git commit(s)
  - a link to the PHP manual entry for the feature

===== References =====

  * https://wiki.php.net/rfc/operator_overloading_gmp
  * https://wiki.php.net/rfc/autoboxing

===== Rejected Features =====

Keep this updated with features that were discussed on the mail lists.