Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:97492
Mailing-List: contact internals-help@lists.php.net; run by ezmlm
Received-SPF: softfail (pb1.pair.com: domain golemon.com does not designate 209.85.218.41 as permitted sender)
MIME-Version: 1.0
Sender: php@golemon.com
In-Reply-To: <CAESVnVoKF7xQKJMjSAO4p0GcYgJy7h=_n04q-VARLM0c9beAwg@mail.gmail.com>
References: <CAESVnVrsf7bdtuAT5+6vGKNmXLr2K3Mnpr=CCnd+o4z9vKMhwQ@mail.gmail.com>
 <CAESVnVpOtMgk0JvWPxwZwoQFBw31cyUvLpRgpgWsZyV_BVHQzg@mail.gmail.com>
 <CAESVnVoc6j2c7MO-_N=1JEe4Z3d3Q-S+UHZ0m-v3TAy+QJfKOg@mail.gmail.com>
 <CAESVnVo5HeOuCwqu=ntgAeTRujOXR=uyKV9nPMtYZ-3z_3_Sxw@mail.gmail.com> <CAESVnVoKF7xQKJMjSAO4p0GcYgJy7h=_n04q-VARLM0c9beAwg@mail.gmail.com>
Date: Thu, 29 Dec 2016 18:32:20 -0800
Message-ID: <CAESVnVq7q3_-fJtZwTs0FRVeONnUnqcaDw+PPgbFGYg_5tF1iA@mail.gmail.com>
To: PHP internals <internals@lists.php.net>
Content-Type: text/plain; charset=UTF-8
Subject: Fwd: [RFC] HashContext as Object
From: pollita@php.net (Sara Golemon)

On Mon, Dec 26, 2016 at 7:35 PM, Sara Golemon wrote:
> I was trawling through Pull Requests and found #660 which I think is a
> nice idea and deserves some attention.  It involves minor BC however,
> so I've updated the patch and presented it for your discussing
> pleasure.
>
Specifically, I should say I'm eager to hear thoughts on the item
listed as an Open Issue.

That is: Currently hash_final() destroys the existing HashContext
resource immediately.  No hope of continuing or even just reissuing
the result.

I'm quite tempted atm, to go the "copy the context, finalize to get a
digest, then restoring to the saved context state".  For most use
cases of this API this is a slight overhead add since the context is
being duplicated then shortly thereafter discarded. (Bear in mind
though, that the hash_init->hash_update->hash_final API is the edge
case already compared to the all-in-one hash() and hash_hmac()
functions).

What doing this does offer, however, is a pathway to reasonable APIs
for streaming hash functions (see SHA3/Keccak Sponge Functions).  As a
minor side benefit, we also get an idempotent hash_final(). So even
though there's a VERY minor BC break involved, I think it's the right
thing to do.  If others agree, I'll update the diff to reflect.

-Sara