Everything

php

PHP_FLOAT_MIN is positive

php.internals

Benjamin Morel

7 years ago

Hi internals,

I just used PHP_FLOAT_MIN for the first time, and was surprised that it is
the smallest **positive** number representable. Is this expected?

This is unlike PHP_INT_MIN, which is the absolute smallest representable
integer, and as such is negative:

echo PHP_INT_MIN; // -9223372036854775808
echo PHP_FLOAT_MIN; // 2.2250738585072E-308

If it is intended, maybe the doc
<https://www.php.net/manual/en/reserved.constants.php> should be clear
about this, at the moment it is just:

Smallest representable floating point number.


Which is confusing IMO.

Cheers,
Ben

Diogo Neves

7 years ago

It really don't make much sense:

<?php

var_dump( PHP_FLOAT_MIN < 0 );
var_dump( PHP_INT_MIN < 0 );

On Wed, Apr 3, 2019 at 10:52 AM Benjamin Morel <benjamin.morel@gmail.com>
wrote:

Joe Watkins

7 years ago

2.2250738585072E-308

This is negative.

Cheers
Joe

On Wed, 3 Apr 2019 at 12:27, Diogo Neves <dafneves@gmail.com> wrote:

Côme Chilliet

7 years ago

Le mercredi 3 avril 2019, 13:47:55 CEST Joe Watkins a écrit :

> 2.2250738585072E-308
>
> This is negative.
>
> Cheers
> Joe

No it’s not.
It’s positive and close to 0.

Côme

Michael Wallner

7 years ago

On 03/04/2019 12:27, Diogo Neves wrote:

> It really don't make much sense:
>
> <?php
>
> var_dump( PHP_FLOAT_MIN < 0 );
> var_dump( PHP_INT_MIN < 0 );
>
> On Wed, Apr 3, 2019 at 10:52 AM Benjamin Morel <benjamin.morel@gmail.com>
> wrote:
>
>> Hi internals,
>>
>> I just used PHP_FLOAT_MIN for the first time, and was surprised that it is
>> the smallest **positive** number representable. Is this expected?
>>
>> This is unlike PHP_INT_MIN, which is the absolute smallest representable
>> integer, and as such is negative:
>>
>> echo PHP_INT_MIN; // -9223372036854775808
>> echo PHP_FLOAT_MIN; // 2.2250738585072E-308
>>
>> If it is intended, maybe the doc
>> <https://www.php.net/manual/en/reserved.constants.php> should be clear
>> about this, at the moment it is just:
>>
>> Smallest representable floating point number.
>>
>>
>> Which is confusing IMO.

FLT_MIN is the smallest real near 0. Are you looking for -FLT_MAX?
With IEEE754 floats, there are positive and negative zeros, so the range
above and below 0 is the same.

--
Regards,
Mike

Benjamin Morel

7 years ago

>
> It really don't make much sense:
> var_dump( PHP_FLOAT_MIN < 0 );
> var_dump( PHP_INT_MIN < 0 );

I quite agree that this is an inconsistency, but I guess that it's here to
stay now for BC reasons. What we can do is fix the doc.

> 2.2250738585072E-308
> This is negative.

It isn't, it is a strictly positive number with a negative exponent:
0.000(...)00022250738585072.

FLT_MIN is the smallest real near 0. Are you looking for -FLT_MAX?

> With IEEE754 floats, there are positive and negative zeros, so the range
> above and below 0 is the same.

It's interesting to know that - PHP_FLOAT_MAX is actually the smallest
representable floating point number (unlike - PHP_INT_MAX). It would be
great to document this as well.

I would propose to change the current documentation to:

PHP_FLOAT_MIN (float)

> Smallest representable POSITIVE floating point number. If you need the
> smallest representable floating point number, use - PHP_FLOAT_MAX.
> Available as of PHP 7.2.0.

What we could also do is introduce another constant to represent the actual
smallest possible float, but its naming would likely conflict with
PHP_FLOAT_MIN, and it does not bring much value, so I'm not sure whether
this is a good idea.
We did introduce a PHP_INT_MIN though, even though it was available as ~
PHP_INT_MAX, so it's worth mentioning.

Thoughts?



On Wed, 3 Apr 2019 at 13:50, Michael Wallner <mike@php.net> wrote:

Rowan Collins

7 years ago

On Wed, 3 Apr 2019 at 13:33, Benjamin Morel <benjamin.morel@gmail.com>
wrote:

> PHP_FLOAT_MIN (float)
> > Smallest representable POSITIVE floating point number. If you need the
> > smallest representable floating point number, use - PHP_FLOAT_MAX.
> > Available as of PHP 7.2.0.
>

I'd avoid the word "smallest". PHP_FLOAT_MIN could be described as having
the "smallest magnitude", and -PHP_FLOAT as being "the largest magnitude,
but negative".

Perhaps:

> Closest representable positive floating point number to zero. If you need

the

> negative number furthest from zero, use -PHP_FLOAT_MAX.

Regards,

--
Rowan Collins
[IMSoP]

Joe Watkins

7 years ago

Thanks for explain.

Cheers
Joe

On Wed, 3 Apr 2019 at 14:40, Rowan Collins <rowan.collins@gmail.com> wrote:

Claude Pache

7 years ago

> Le 3 avr. 2019 à 11:51, Benjamin Morel <benjamin.morel@gmail.com> a écrit :
>
> Hi internals,
>
> I just used PHP_FLOAT_MIN for the first time, and was surprised that it is
> the smallest **positive** number representable. Is this expected?
>
> This is unlike PHP_INT_MIN, which is the absolute smallest representable
> integer, and as such is negative:
>
> echo PHP_INT_MIN; // -9223372036854775808
> echo PHP_FLOAT_MIN; // 2.2250738585072E-308
>
> If it is intended, maybe the doc
> <https://www.php.net/manual/en/reserved.constants.php> should be clear
> about this, at the moment it is just:
>
> Smallest representable floating point number.
>
>
> Which is confusing IMO.
>
> Cheers,
> Ben

And in order to avoid further confusion, the constant ought to be renamed PHP_FLOAT_MIN_POSITIVE.

Rust named it correctly: https://doc.rust-lang.org/std/f64/constant.MIN_POSITIVE.html <https://doc.rust-lang.org/std/f64/constant.MIN_POSITIVE.html>

—Claude

Sara Golemon

7 years ago

On Wed, Apr 3, 2019 at 4:52 AM Benjamin Morel <benjamin.morel@gmail.com>
wrote:

> I just used PHP_FLOAT_MIN for the first time, and was surprised that it is
> the smallest **positive** number representable. Is this expected?
>
> This is unlike PHP_INT_MIN, which is the absolute smallest representable
> integer, and as such is negative:
>
> echo PHP_INT_MIN; // -9223372036854775808
> echo PHP_FLOAT_MIN; // 2.2250738585072E-308
>
> If it is intended, maybe the doc
> <https://www.php.net/manual/en/reserved.constants.php> should be clear
> about this, at the moment it is just:
>
> Smallest representable floating point number.
>
>
> Which is confusing IMO.
>
> Perhaps confusing, but not without precedent.  See the C++ definitions for

::min() and ::lowest() on
https://en.cppreference.com/w/cpp/types/numeric_limits.

std::numeric_limits<double>::min() is a positive value very near zero.
std::numeric::limits<double>::lowest() is a very large negative value.

-Sara