Implement <=> (spaceship/comparison) operator

[hikari-no-yume]

Implementation of RFC: https://wiki.php.net/rfc/combined-comparison-operator

Also see Implement cmp() function.

[hikari-no-yume]

ext/tokenizer data would need regenerating, but the shell script is completely broken on my machine so someone else would need to do it. I thought I needed GNU awk, but that didn't fix it.

[smalyshev]

See also: https://wiki.php.net/rfc/combined-comparison-operator

[hikari-no-yume]

Huh, I had no idea there was prior art RFC-wise. Maybe @dshafik would let me take over that RFC.

I like his token name: T_ENTERPRISE.

[smalyshev]

It would be nice to add some objects - like GMP - here.

[hikari-no-yume]

Fair point. This is based on the > test, so the others should be updated if that's the case.

[laruence]

@TazeTSchnitzel okey. you are right... I am in a insane status now.

[r3wt]

what point would this even serve? lol

[hikari-no-yume]

@r3wt It makes it easier to write comparison functions used in sorting, for example.

[Ragazzo]

@TazeTSchnitzel great , it is really useful, especially when building neural networks

[acasademont]

Hi Andrea! Why not use (!$failed) or $failed === false? Strict comparisons are nicer if possible. Thanks!

[hikari-no-yume]

Ah, it's like that because this is based on either the >= or the <= test (can't remember). It wasn't a deliberate choice on my part.

I could fix it, but I don't see the point, especially as it'd just make this inconsistent with the other tests.

[acasademont]

Yeah, I guess those tests date back to a time when strict comparisons where not that popular! And thanks for the hard work :)

[vstinner]

Hi,

Sorry, I don't know if it's the best place to comment the RFC https://wiki.php.net/rfc/combined-comparison-operator

I'm surprised by PHP adds <=> operator in PHP 7, while Python 3 dropped cmp(). cmp() was dropped because it had many issues. Example:
https://stackoverflow.com/questions/20202418/why-is-the-cmp-parameter-removed-from-sort-sorted-in-python3-0

[hikari-no-yume]

@Haypo from what I can see, Python's removal of __cmp__ has no relevance to PHP.

[vstinner]

@Haypo from what I can see, Python's removal of cmp has no relevance to PHP.

PHP has also objects which have surprising behaviour on comparison. For example, float not a number (NaN). Try with "$nan = acos(8);".

$ php -a Interactive shell php > $a=acos(8); php > $b=acos(10); php > var_dump($a); float(NAN) php > var_dump($b); float(NAN) php > var_dump($a > $b); bool(false) php > var_dump($a >= $b); bool(false) php > var_dump($a < $b); bool(false) php > var_dump($a <= $b); bool(false) php > var_dump($a == $b); bool(false) php > var_dump($a != $b); bool(true) 

What should be the result of $a <=> $b for NaN?

See also:

• Get rid of more references to cmp: http://bugs.python.org/issue1717
• Remove cmp parameter to list.sort() and builtin.sorted(): http://bugs.python.org/issue1771

[hikari-no-yume]

Those aren't objects.

[hikari-no-yume]

@smalyshev I just rebased the branch onto master and squashed the commits, FYI. (This pull request is dead, though.)

[Ragazzo]

@TazeTSchnitzel why is it closed and dead is you said , it is a great feature ?

[hikari-no-yume]

@Ragazzo I quit PHP development. However, @smalyshev revived it, and it passed.

[coveralls]

Changes Unknown when pulling d8b7a44 on TazeTSchnitzel:spaceship_operator into * on php:master*.

[hikari-no-yume]

why is @coveralls bothering me about this. the pull request is closed.

[dshafik]

@TazeTSchnitzel asking the right questions. I got the email too.