I think you've misunderstood my post, so I'll start again.
>From Zend_Locale_Math_PhpMath::Pow():
$result = pow($op1, $op2);
pow() is a function built into PHP. It does not work on "BigIntegers" of
infinite precision. This is in contrast to bcmath.
A demonstration:
<?php
require_once 'Zend/Locale/Math.php';
require_once 'Zend/Locale/Math/PhpMath.php';
Zend_Locale_Math_PhpMath::disable();
$a = Zend_Locale_Math::Pow('2', '128');
$b = Zend_Locale_Math_PhpMath::Pow('2', '128');
var_dump($a);
echo "\r\n<br />\r\n";
var_dump($b);
?>
What do you get as the output? Here's what I get:
string(39) "340282366920938463463374607431768211456"
string(19) "3.4028236692094E+38"
Those are very definitely not equal. Tell me - how do you propose inferring
the least significant digit from $b? It can be determined from $a by doing
$a[strlen($a) - 1]. What are you going to do for $b?
The fact is that when bcmath is unavailable, Zend_Locale_Math does *not*
behave identically to how it does when bcmath is available. To say the
fallback solution behaves identically to the normal behavior is, frankly,
incorrect.
Here's the original BigInteger library I linked to:
http://phpseclib.cvs.sourceforge.net/viewvc/*checkout*/phpseclib/phpseclib/Math/BigInteger.php?revision=1.5
That gives you the *same* result when gmp is available, when bcmath is
available, and when neither are available. That's the behavior I would
expect from a BigInteger class that presumes to have a PHP fallback solution
- not this "we'll give you completely different output" approach that is
being taken with Zend.
thomasW wrote:
>
> See my notes below
>
> Greetings
> Thomas Weidner, I18N Team Leader, Zend Framework
> http://www.thomasweidner.com
>
> ----- Original Message -----
> From: "zelnaga" <[email protected]>
> To: <[email protected]>
> Sent: Wednesday, February 25, 2009 9:37 PM
> Subject: Re: [fw-general] pure-PHP BigInteger support
>
>
>>
>>
>> thomasW wrote:
>>>
>>> There is already a pure PHP solution within ZF which can be used.
>>>
>>> Zend_Locale_Math is a bcmath clone which does automatically use BCMath
>>> and
>>> when it's not available switch to pure PHP implementation. For examples
>>> on
>>> usage look at Zend_Locale_Format or Zend_Measure_Abstract.
>>
>> http://framework.zend.com/svn/framework/standard/trunk/library/Zend/Locale/Math/PhpMath.php
>>
>>
>> Interesting. Correct me if I'm wrong, but it appears to me that when
>> bcmath
>> is unavailable that library ceases to be a multiple precision integer
>> arithmetic library (ie. BigInteger) and instead supports regular old
>> 32-bit
>> integers (or 64-bit for 64-bit machines)?
>
> No, it does not limit to 32 bit integer. Still there is a sort of
> limitation
> which is related to the usecase this lib was created for.
>
>> If so, I'm not even sure it'd be
>> accurate to say that it "falls back" on anything - if bcmath is
>> unavailable
>> its feature set becomes completely different.
>
> No it does not.
>
>> If you absolutely need
>> 128-bit integers and bcmath is unavailable that library isn't going to be
>> sufficient
>
> We're calculating dates with a unixtimestamp from BC (-4000) which extends
> 32bit with ease.
>
>> and if you have bcmath and can live with 32-bit integers that
>> library is probably going to slow things down for you
>
> Why should it ?
> When BCMath is available you will get the result from BCMath and not the
> lib.
>
>> (bcadd(), I suspect, can't compare in speed to the + operator).
>
> You have to decide if you are in need of BCMath or of PHP limited integer.
> This is a design decision and not a fault or problem of a class.
> You are ALWAYS slower when you have to use a component or extension than
> using PHP native "+" notation.
>
>> If that library actually does support integers of any length regardless
>> of
>> whether or not bcmath is installed then shouldn't that library - and not
>> Zend_Crypt_Math_BigInteger - be used by Zend_Crypt_DiffieHellman?
>
> Can't say anything about that. Zend_Crypt is no longer developed by it's
> maintainer and unchanged in incubator since mid 2008.
> So I expect that this component is dead.
>
>> --
>> View this message in context:
>> http://www.nabble.com/pure-PHP-BigInteger-support-tp22068811p22211234.html
>> Sent from the Zend Framework mailing list archive at Nabble.com.
>
>
>
--
View this message in context:
http://www.nabble.com/pure-PHP-BigInteger-support-tp22068811p22213096.html
Sent from the Zend Framework mailing list archive at Nabble.com.
