Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:103012
Mailing-List: contact internals-help@lists.php.net; run by ezmlm
Received-SPF: pass (pb1.pair.com: domain gmail.com designates 209.85.167.44 as permitted sender)
MIME-Version: 1.0
References: <CANA3DesUZ1T=vfvgb7T7CqMqF3KvrMzykFEAS5Y0y6Sxo0c0_A@mail.gmail.com>
In-Reply-To: <CANA3DesUZ1T=vfvgb7T7CqMqF3KvrMzykFEAS5Y0y6Sxo0c0_A@mail.gmail.com>
Date: Wed, 1 Aug 2018 13:12:31 -0600
Message-ID: <CAFMT4NrHR97_0JxHerS9t6FDLHEe0+vVMTZgskiDM29YWFuLPg@mail.gmail.com>
To: marcospassos.com@gmail.com
Cc: internals <internals@lists.php.net>
Content-Type: text/plain; charset="UTF-8"
Subject: Re: [PHP-DEV] Array max size
From: levim@php.net (Levi Morrison)

On Wed, Aug 1, 2018 at 12:42 PM Marcos Passos
<marcospassos.com@gmail.com> wrote:
>
> Whenever you look for more information about the maximum size of an array,
> you find someone saying that "PHP arrays do not have a maximum size, but
> the amount of memory available". However, I could not find any excerpt in
> PHP documentation that supports that.
>
> Even if the engine imposes no hard limit on the size of an array, in fact,
> there an inherent limit that is assumed by some functions, and that is what
> I would like to suggest making explicit on the documentation. The lack of
> this definition leads to inconsistencies and leaves several open questions,
> including:
>
>    - If no limit exists, then it's conceptually possible to have an array
>    with *PHP_INT_MAX + 1* elements. In that sense, what would be the return
>    of the *\count()*?
>    - The function *range* validates the size of the resulting range against
>    the maximum size of the hash table (defined internally as *HT_MAX_SIZE*),
>    and throw an error if it exceeds that value. Is that the limit?
>    - he function *array_fill*, in contrast, does not rely on *HT_MAX_SIZE*
>    for validating the size of the result. But, why?
>    - The documentation says that omitting the third parameter of
>    array_split is equivalent to \count($array). If the maximum number
>    representable in PHP is *PHP_INT_MAX*, is the max size the same as
>    *PHP_INT_MAX*?
>
> There are other examples, but I think these are enough to make the point.
>
> My understanding is that the conceptual limit is *PHP_INT_MAX*, as there is
> no way to represent the size above this value. If so, the interval that
> represents all possibles indexes for an array is defined as *0 <= index <=
> PHP_INT_MAX -1*. That definition aligns with all functions that support
> negative length as *-PHP_INT_MAX* is equivalent to the start of the array.
>
> Could you guys please share your thoughts on this topic?

If we look at the definition of _zend_array:

  https://github.com/php/php-src/blob/1a303f1a2b6ed8853fea3a2079999eef7f462d86/Zend/zend_types.h#L237-L257

It stores element counts, num used, etc, in uint32_t. It seems to me
that therefore they can't hold more than UINT32_MAX number of
elements.