Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:99101
Mailing-List: contact internals-help@lists.php.net; run by ezmlm
Received-SPF: pass (pb1.pair.com: domain gmail.com designates 209.85.216.174 as permitted sender)
MIME-Version: 1.0
In-Reply-To: <94cdf88b-9933-a72d-d351-546da495d4ab@beccati.com>
References: <CADxrgCMNFqLAoMCbssqkajgn9=_7HTbk70LybwpD+Co3aQzmJg@mail.gmail.com>
 <94cdf88b-9933-a72d-d351-546da495d4ab@beccati.com>
Date: Thu, 18 May 2017 12:16:58 +0200
Message-ID: <CADxrgCNOKw17xRQrDBJaHo7qmf9w++U1EH5K8z2dyhauVceDig@mail.gmail.com>
To: Matteo Beccati <php@beccati.com>
Cc: "internals@lists.php.net" <internals@lists.php.net>
Content-Type: multipart/alternative; boundary="001a1141058e8864ec054fc9b477"
Subject: Re: [PHP-DEV] [RFC] deprecate PDO::PARAM_NULL
From: adam.baratz@gmail.com (Adam Baratz)

--001a1141058e8864ec054fc9b477
Content-Type: text/plain; charset="UTF-8"

>
> "A handful of people on the internals mail list have asked for it to be
> removed" - where and when?
>

I didn't mean this as a justification or as any kind of vote, just as a
prompt for a formal discussion.

"Supporting it adds some weight to PDO" what kind of weight? How many
> lines / microseconds are we going to save?
>

It would mostly make some of the driver code easier to follow. There's some
slightly strange logic in places, to account for being able to bind NULL by
type or value. There are 14 references to PDO_PARAM_NULL across pdo,
pdo_mysql, and pdo_pgsql. There are 6 references to the type in unit tests.

I would agree that NULL is not a type and the constant shouldn't have
> been there in the first place, but I think it could be a bit too late now.
>
> The proposal itself is rather evil: it would break lots of code (a quick
> github search shows 95k+ results) for basically no gain. Or maybe I just
> didn't get it.


People could future-proof their code with PDO as it exists today. I'd want
to deprecate it with the same schedule we've used to remove whole
extensions. Is there a schedule that would make you comfortable with the
change?

Most of the gain would be in cleaning up the PDO type system. The more it
resembles the SQL type system, the easier it will be for statements to be
portable between drivers. Having a NULL type just feels like an anomaly
that should be corrected.

In all honesty, I'm ambivalent about this issue. It's certainly not causing
any problems. But if enough people feel it should be removed, I'm happy to
do the work to remove it.

Thanks,
Adam

--001a1141058e8864ec054fc9b477--