Object comparison View as Markdown

Hi!

Doc bug? Or code bug? I'm inclined to call it a code bug, but wanted
others' thoughts.

I would say comparing object to a number makes little sense, so no
reason to define any specific result there. It may be true, false or
bologna sandwich.
The docs say what happens when the first parameter is object, but say
nothing what happens if the second one is object. This is for reason -
if you compare object to array, they can't both be greater, something
has to take priority. The docs say first arg takes priority. So by docs,
comparison of (number, object) has no defined value, while comparison of
(object, number) has.

Stanislav Malyshev, Software Architect
SugarCRM: http://www.sugarcrm.com/
(408)454-6900 ext. 227

From: http://php.net/manual/en/language.operators.comparison.php

An object compared to anything which is not a bool, null, or object
should result in the object appearing to be greater than the other
operand. For example:

$a = new stdClass();
$b = new stdClass();

var_dump(null < $a);
var_dump(false < $a);
var_dump(true == $a);
var_dump($a == $b);
var_dump(0 < $a);
var_dump(1 < $a); // false
var_dump(2 < $a); // false
var_dump("foo" < $a);
var_dump("2" < $a);
var_dump(tmpfile() < $a);

Based on docs, I expect all nine of these to yield true, however in
practice, the two marked "false" come out as false because the RHS
object is converted to an integer (1), contrary to the docs.

Doc bug? Or code bug? I'm inclined to call it a code bug, but wanted
others' thoughts.

Hi Sara,

I believe the documentation is failing us here.

From what I see the docs are pretty lacking in what should be expected
in terms of object comparison.

For example, the page at http://php.net/types.comparisons doesn't even
include objects in the comparison table.

Of course by looking at the "Comparison with Various Types" table at
http://php.net/language.operators.comparison one gets the impression
that the "Operand 1" and "Operand 2" columns are to signify LVALUE and
RVALUE operands, respectively. Obviously this isn't the case and the
documentation just fails us here.

var_dump(new stdclass < 1, newstdcalss > 1, 1 < new stdclass, 1 > new
stdclass); // false, false, false, false

Clearly there are cases when the object can be neither less-than nor
greater-than an operand in a comparison.

The statement "object is always greater" in that table is misleading
and doesn't tell us the whole truth.

For example,
class foo { public function __toString() { return ''; } }
var_dump(new foo < '', new foo > '', '' < new foo, '' > new foo); //
false, false, false, false

Obviously there are more complex cases where the object may not pan
out to be greater than a non-object type that the documentation fails
to address.

-Sara

From: http://php.net/manual/en/language.operators.comparison.php

An object compared to anything which is not a bool, null, or object
should result in the object appearing to be greater than the other
operand. For example:

$a = new stdClass();
$b = new stdClass();

var_dump(null < $a);
var_dump(false < $a);
var_dump(true == $a);
var_dump($a == $b);
var_dump(0 < $a);
var_dump(1 < $a); // false
var_dump(2 < $a); // false
var_dump("foo" < $a);
var_dump("2" < $a);
var_dump(tmpfile() < $a);

Based on docs, I expect all nine of these to yield true, however in
practice, the two marked "false" come out as false because the RHS
object is converted to an integer (1), contrary to the docs.

Doc bug? Or code bug? I'm inclined to call it a code bug, but wanted
others' thoughts.

-Sara

You seem to be reporting a bug ("contrary to the docs"), but you might
be suggesting that the doc be corrected (since doc and behaviour should
always match).

I would suggest being very careful about implementing something as
central as a change to how comparisons are made, that ought to be
approached with great caution because it has the potential for breaking
a huge amount of existing code.

[It does occur to me however, simply as a point of interest, that
objects traditionally have both a constructor and a destructor, and the
idea of objects also having a comparator makes a certain amount of
sense; a default object comparator might implement the currently defined
comparisons while allowing derived classes to override the default
behaviour and leaving comparisons not involving objects undisturbed.]

Comparing objects to scalars is really an apples/oranges comparision to
begin with.

I would like to place a suggestion for comparing objects (I hope it is no problem, because this does not have anything to do with Sara's question - but it came to my mind when I read her mail). It would be a great feature if objects could be compared to other objects with ">", "<" and the other operators, like it is suggested in RFC https://wiki.php.net/rfc/comparable
The DateTime class offers this feature - it would be nice if this could be made usable for userland classes/objects, too.

Best regards

Christian Stoller

-----Original Message-----
From: php@golemon.com [mailto:php@golemon.com] On Behalf Of Sara Golemon
Sent: Friday, November 09, 2012 1:07 AM
To: PHP internals
Subject: [PHP-DEV] Object comparison

From: http://php.net/manual/en/language.operators.comparison.php

An object compared to anything which is not a bool, null, or object
should result in the object appearing to be greater than the other
operand. For example:

$a = new stdClass();
$b = new stdClass();

var_dump(null < $a);
var_dump(false < $a);
var_dump(true == $a);
var_dump($a == $b);
var_dump(0 < $a);
var_dump(1 < $a); // false
var_dump(2 < $a); // false
var_dump("foo" < $a);
var_dump("2" < $a);
var_dump(tmpfile() < $a);

Based on docs, I expect all nine of these to yield true, however in
practice, the two marked "false" come out as false because the RHS
object is converted to an integer (1), contrary to the docs.

Doc bug? Or code bug? I'm inclined to call it a code bug, but wanted
others' thoughts.

-Sara

hi!

From: http://php.net/manual/en/language.operators.comparison.php

An object compared to anything which is not a bool, null, or object
should result in the object appearing to be greater than the other
operand. For example:

$a = new stdClass();
$b = new stdClass();

var_dump(null < $a);
var_dump(false < $a);
var_dump(true == $a);
var_dump($a == $b);
var_dump(0 < $a);
var_dump(1 < $a); // false
var_dump(2 < $a); // false
var_dump("foo" < $a);
var_dump("2" < $a);
var_dump(tmpfile() < $a);

Based on docs, I expect all nine of these to yield true, however in
practice, the two marked "false" come out as false because the RHS
object is converted to an integer (1), contrary to the docs.

Doc bug? Or code bug? I'm inclined to call it a code bug, but wanted
others' thoughts.

As stated by other before, comparing scalars and objects sound wrong
in the 1st place. But I have seen way too many weird codes relying on
weird things :). That's why I would not be in favor of changing
anything in this area as some code out there may rely on that (have
the feeling that object comparison has been like that since 5.0).

Cheers,

Pierre

@pierrejoye | http://blog.thepimp.net | http://www.libgd.org