Everything

php

Home » Project » Newsgroups » php.internals » foreach with null »

foreach with null

php.internals

Marcus Börger

20 years ago

Hello internals,

i would like to change foreach a tiny bit. Actually i don't like the
warning when using foreach with null.

marcus@zaphod /usr/src/php-cvs $ php -r '$n=NULL; foreach($n as $v);'
make: `sapi/cli/php' is up to date.

Warning: Invalid argument supplied for foreach() in Command line code on line 1
marcus@zaphod /usr/src/php-cvs $ php -r 'foreach(NULL as $v);'
make: `sapi/cli/php' is up to date.

Warning: Invalid argument supplied for foreach() in Command line code on line 1

The way we use NULL in PHP should result in no warning here. The patch for
this can be found here:
http://php.net/~helly/php/ext/ze2/ze2-foreach-null-20051025.diff.txt
It was made with head but applies to 5.1 as well.

Any opinions?


Best regards,
Marcus

Rob Richards

20 years ago

Marcus Boerger wrote:

>  i would like to change foreach a tiny bit. Actually i don't like the
>warning when using foreach with null.
>
>

Marcus, you already know my feelings on this one.
+1

Rob

Antony Dovgal

20 years ago

I'd say silently ignoring it and moving along is fine here.
Only if the var IS_NULL, but this is the case.

So I'm +1.

On 25.10.2005 02:22, Marcus Boerger wrote:

> Hello internals,
>
>   i would like to change foreach a tiny bit. Actually i don't like the
> warning when using foreach with null.
>
> marcus@zaphod /usr/src/php-cvs $ php -r '$n=NULL; foreach($n as $v);'
> make: `sapi/cli/php' is up to date.
>
> Warning: Invalid argument supplied for foreach() in Command line code on line 1
> marcus@zaphod /usr/src/php-cvs $ php -r 'foreach(NULL as $v);'
> make: `sapi/cli/php' is up to date.
>
> Warning: Invalid argument supplied for foreach() in Command line code on line 1
>
> The way we use NULL in PHP should result in no warning here. The patch for
> this can be found here:
> http://php.net/~helly/php/ext/ze2/ze2-foreach-null-20051025.diff.txt
> It was made with head but applies to 5.1 as well.
>
> Any opinions?

--
Wbr,
Antony Dovgal

John Coggeshall

20 years ago

+1 as well if and only if this applies to IS_NULL

On Tue, 2005-10-25 at 02:45 +0400, Antony Dovgal wrote:

Jani Taskinen

20 years ago

Will the patch make it silently ignore also empty string and 0?
If not, -1. (what's with null being suddenly equal to empty array? :)

--Jani

On Tue, 25 Oct 2005, Marcus Boerger wrote:

>
> Hello internals,
>
>  i would like to change foreach a tiny bit. Actually i don't like the
> warning when using foreach with null.
>
> marcus@zaphod /usr/src/php-cvs $ php -r '$n=NULL; foreach($n as $v);'
> make: `sapi/cli/php' is up to date.
>
> Warning: Invalid argument supplied for foreach() in Command line code on line 1
> marcus@zaphod /usr/src/php-cvs $ php -r 'foreach(NULL as $v);'
> make: `sapi/cli/php' is up to date.
>
> Warning: Invalid argument supplied for foreach() in Command line code on line 1
>
> The way we use NULL in PHP should result in no warning here. The patch for
> this can be found here:
> http://php.net/~helly/php/ext/ze2/ze2-foreach-null-20051025.diff.txt
> It was made with head but applies to 5.1 as well.
>
> Any opinions?
>
>
> Best regards,
> Marcus
>
>

--
Give me your money at @ <http://pecl.php.net/wishlist.php/sniper>
Donating money may make me happier and friendlier for a limited period!
Death to all 4 letter abbreviations starting with P!

Ants Aasma

20 years ago

It currently is equal:
$ php -r 'var_dump(null == array()); var_dump((array) null);'
bool(true)
array(0) {
}

Ants

Jani Taskinen wrote:

Antony Dovgal

20 years ago

On 25.10.2005 19:39, Ants Aasma wrote:

> It currently is equal:
> $ php -r 'var_dump(null == array()); var_dump((array) null);'
> bool(true)
> array(0) {
> }

No, there aren't equal.
Use === and you'll see the difference.

--
Wbr,
Antony Dovgal

Ants Aasma

20 years ago

Antony Dovgal wrote:

> On 25.10.2005 19:39, Ants Aasma wrote:
>
>> It currently is equal:
>> $ php -r 'var_dump(null == array()); var_dump((array) null);'
>> bool(true)
>> array(0) {
>> }
>
> No, there aren't equal.
> Use === and you'll see the difference.

At least the manual says that the == operator is named equal and the ===
operator is named identical. Actually I wanted to point out that null is
a bit of a special case in PHP:
$ php -r 'var_dump(null == 0); var_dump(null == array()); var_dump(0 ==
array());'
bool(true)
bool(true)
bool(false)
$ php -r 'var_dump((array) ""); var_dump((array) 0); var_dump((array)
null);'
array(1) {
[0]=>
string(0) ""
}
array(1) {
[0]=>
int(0)
}
array(0) {
}

+-0 on the foreach issue, but with the SPL Iterators I can see where
Marcus is coming from.

Ants

Christian Schneider

20 years ago

Ants Aasma wrote:

> +-0 on the foreach issue, but with the SPL Iterators I can see where
> Marcus is coming from.

Ok, I can see the reason behind it now as casting to array is not an
option for objects with SPL Iterators. One little side-note: Programmers
will still have to do an if($obj) check if they use $obj for just about
anything but foreach.

After thinking it over you convinced me that treating null (and only
null) like an empty array in foreach() might be a good idea, probably
even for the non-SPL case.

- Chris

Marcus Börger

20 years ago

Hello Christian,

Tuesday, October 25, 2005, 7:28:09 PM, you wrote:

> Ants Aasma wrote:
>> +-0 on the foreach issue, but with the SPL Iterators I can see where
>> Marcus is coming from.

> Ok, I can see the reason behind it now as casting to array is not an
> option for objects with SPL Iterators. One little side-note: Programmers
> will still have to do an if($obj) check if they use $obj for just about
> anything but foreach.

> After thinking it over you convinced me that treating null (and only
> null) like an empty array in foreach() might be a good idea, probably
> even for the non-SPL case.

Hey you two guys pointed out the reason very exactly! Only the usecase
wasn't SPL but a potential change i want to do in SimpleXML and the reson
SXE behaves like it currently does. But yes SPL and PDO would probably
benefit in the same way.

Best regards,
Marcus

Christian Schneider

20 years ago

Marcus Boerger wrote:

>   i would like to change foreach a tiny bit. Actually i don't like the
> warning when using foreach with null.
>
> The way we use NULL in PHP should result in no warning here. The patch for

Can you elaborate on the "The way we use NULL in PHP" part?

IMPW (In my PHP world) null represents an undefined variable. Trying to
iterate over it could be a programming mistake. Like iterating over a
scalar value. I normally use foreach ((array)$var as $value) if $var is
of "mixed" type which also catches null. But most of the time I'm more
comfortable with making sure I provide a real array in all cases and
don't rely on magic.

Just my two cents,
- Chris

David Zülke

20 years ago

> IMPW (In my PHP world) null represents an undefined variable. Trying to
> iterate over it could be a programming mistake. Like iterating over a
> scalar value. I normally use foreach ((array)$var as $value) if $var is
> of "mixed" type which also catches null. But most of the time I'm more
> comfortable with making sure I provide a real array in all cases and
> don't rely on magic.

Good point. Smells like a -1 to me.

- David

Ilia A.

20 years ago

IMHO this is a bad idea, if you really wanted to support NULL in a
particular case you, as a developer could do:
php -r '$n=NULL; foreach((array)$n as $v);'

Making NULL usage transparent where array is expects will lead to
difficult to debug code.

-1 for removing the warning.

Ilia

Jochem Maas

20 years ago

Ilia Alshanetsky wrote:

> IMHO this is a bad idea, if you really wanted to support NULL in a
> particular case you, as a developer could do:
> php -r '$n=NULL; foreach((array)$n as $v);'
>
> Making NULL usage transparent where array is expects will lead to
> difficult to debug code.

hi :-)

the warning for foreach() in this case seems to be inconsistent with:

php -r '$a = null; $b = array(); var_dump((empty($a) === empty($b)), (count($a) === count($b)));'

which gives no warnings or otherwise, also as a lesser developer than you I don't
buy the notion that dropping the warning makes debugging more difficult because
advanced code should be checking vars properly before using them and, well,
for any given foreach loop there is often enough an 'if (!count($array))'-like
statement to take care of the 'empty' case. and in cases when speed is very important
it saves an explicit cast (is that more expensive anyway?) and it means not requiring
an '@' to suppress which I do know is more expensive than not having any E_* to suppress at all.

on the face of it I understand begin against, feedback is good.
but removing the warning would make the engine behaviour just a little
more consistent - which is nice, no?

<sidenote>
did anyone remember that php5beta used to allow:

function(MyCLass $x = null) {}

which was nice from the start fom a user perspective,
I was puzzled as to why it went, glad it came back and
hope it says :-)
</sidenote>

thanks for php, and kind regards in general,
jochem

Jani Taskinen

20 years ago

On Thu, 27 Oct 2005, Jochem Maas wrote:

> advanced code should be checking vars properly before using them and, well,
> for any given foreach loop there is often enough an 'if (!count($array))'-like
> statement to take care of the 'empty' case. and in cases when speed is very
> important

..and where count() magically returns 1 f.e. for FALSE. :)
So really you can't rely on that. :)

--Jani

Jochem Maas

20 years ago

Jani Taskinen wrote:

> On Thu, 27 Oct 2005, Jochem Maas wrote:
>
>> advanced code should be checking vars properly before using them and,
>> well,
>> for any given foreach loop there is often enough an 'if
>> (!count($array))'-like
>> statement to take care of the 'empty' case. and in cases when speed is
>> very important
>
>
>     ..and where count() magically returns 1 f.e. for FALSE. :)
>     So really you can't rely on that. :)

true - although personally I always do if(is_array($r) && count($r)) in such cases.
so it seems that there is inconsistency regardless of
whether the warning is removed or not.

out of interest what are the ramifications for SPL and simpleXML
if the warning is not dropped?

Marcus Börger

20 years ago

Hello Jochem,

Thursday, October 27, 2005, 12:49:57 PM, you wrote:

> Jani Taskinen wrote:
>> On Thu, 27 Oct 2005, Jochem Maas wrote:
>>
>>> advanced code should be checking vars properly before using them and,
>>> well,
>>> for any given foreach loop there is often enough an 'if
>>> (!count($array))'-like
>>> statement to take care of the 'empty' case. and in cases when speed is
>>> very important
>>
>>
>>     ..and where count() magically returns 1 f.e. for FALSE. :)
>>     So really you can't rely on that. :)

> true - although personally I always do if(is_array($r) && count($r)) in such cases.
> so it seems that there is inconsistency regardless of
> whether the warning is removed or not.

> out of interest what are the ramifications for SPL and simpleXML
> if the warning is not dropped?

Right now ther is no place in SPL that would benefit. But both SXE and PDO
have to create empty instances to allow code that does not need further
cehecks before sending to foreach(). So the idea is actually some speed
improvement and in case of SimpleXML i'd like to be able to check existance
of a sub element by the following: !is_null($sxe->element)

Best regards,
Marcus

Andi Gutmans

20 years ago

I think we should be keeping the warning (i.e. -1 for the change).
I don't see a reason why overloaded objects can't expose an API of an
empty array() instead of changing this behavior to fit the
implementation. But even if it weren't so, there has to be a good
reason to change the general case.

Andi

Derick Rethans

20 years ago

On Wed, 26 Oct 2005, Ilia Alshanetsky wrote:

> IMHO this is a bad idea, if you really wanted to support NULL in a
> particular case you, as a developer could do:
> php -r '$n=NULL; foreach((array)$n as $v);'
>
> Making NULL usage transparent where array is expects will lead to
> difficult to debug code.
>
> -1 for removing the warning.

Yes, I agree, we should not remove this warning.

Derick

--
Derick Rethans
http://derickrethans.nl | http://ez.no | http://xdebug.org

Hannes Magnusson

20 years ago

On 10/27/05, Derick Rethans <derick@php.net> wrote:

>
> On Wed, 26 Oct 2005, Ilia Alshanetsky wrote:
>
> > IMHO this is a bad idea, if you really wanted to support NULL in a
> > particular case you, as a developer could do:
> > php -r '$n=NULL; foreach((array)$n as $v);'
> >
> > Making NULL usage transparent where array is expects will lead to
> > difficult to debug code.
> >
> > -1 for removing the warning.
>
> Yes, I agree, we should not remove this warning.
>
> Derick

+1 on *not* removing the warning. I can't see any reson why it should be
removed.

php -r '$foo = array(2, 3, 4); echo substr("foobar", $foo). "\n";'
currently prints "oobar". Would like to see, as Jani suggested, E_STRICT
notice thrown when wrong data type is passed to a function/language
construct.

- Hannes

Derick Rethans

20 years ago

On Thu, 27 Oct 2005, Hannes Magnusson wrote:

> On 10/27/05, Derick Rethans <derick@php.net> wrote:
> >
> > On Wed, 26 Oct 2005, Ilia Alshanetsky wrote:
> >
> > > IMHO this is a bad idea, if you really wanted to support NULL in a
> > > particular case you, as a developer could do:
> > > php -r '$n=NULL; foreach((array)$n as $v);'
> > >
> > > Making NULL usage transparent where array is expects will lead to
> > > difficult to debug code.
> > >
> > > -1 for removing the warning.
> >
> > Yes, I agree, we should not remove this warning.
>
> +1 on *not* removing the warning. I can't see any reson why it should be
> removed.

Erm, that's what I meant. +1 with *not* is the same as -1 without....

Derick

--
Derick Rethans
http://derickrethans.nl | http://ez.no | http://xdebug.org