Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:94436
Mailing-List: contact internals-help@lists.php.net; run by ezmlm
Received-SPF: pass (pb1.pair.com: domain gmail.com designates 209.85.161.173 as permitted sender)
MIME-Version: 1.0
In-Reply-To: <7716d180-2365-efc9-e9b3-9fe0998e273b@fleshgrinder.com>
References: <7716d180-2365-efc9-e9b3-9fe0998e273b@fleshgrinder.com>
Date: Fri, 8 Jul 2016 23:54:29 +0200
Message-ID: <CAF+90c_F_YgpsCkfAqZdj2VzeiUGRqUmmguGhHkf-9Tdrvt5Sg@mail.gmail.com>
To: PHP internals <internals@lists.php.net>
Content-Type: multipart/alternative; boundary=94eb2c1461fad53067053726d810
Subject: Re: [PHP-DEV] [RFC] [Vote] var_info
From: nikita.ppv@gmail.com (Nikita Popov)

--94eb2c1461fad53067053726d810
Content-Type: text/plain; charset=UTF-8

On Fri, Jul 8, 2016 at 9:49 PM, Fleshgrinder <php@fleshgrinder.com> wrote:

> Hi!
>
> Voting on var_info has opened and will stay open until 2016-07-22:
>
> https://wiki.php.net/rfc/var_info
>

Not a big fan of this in general (imho belongs in userland), but I'd like
to comment on one specific issue:

You are doing a callability check for all arrays and strings. I wouldn't be
comfortable passing any data that potentially derives from external sources
(aka all data) to that function. E.g. this could trigger autoloading on
user-provided values -- while we have some protections in place to avoid
the most egregious security vulnerabilities, I wouldn't be comfortable with
that. There probably other funky things that can happen here.

Lastly, I'll just leave this list of type-related improvements I'd actually
like to see:
 * Make is_object() return true for IncompleteClass. This is just
ridiculous. I care zero about the theoretical BC impact.
 * Make gettype() return 'resource (unknown type)'  instead of just
'unknown type' for closed resources. This makes the output clear while
still leaving the closed resource distinction.
 * Normalize error message. IIRC currently our type error messages are
really weird, in that they use terms like "integer" which are not actually
valid type hints (or have different meaning). Fixing this doesn't even need
a proposal, just a PR.

Regards,
Nikita

--94eb2c1461fad53067053726d810--