Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:100198
Mailing-List: contact internals-help@lists.php.net; run by ezmlm
Received-SPF: error (pb1.pair.com: domain kelunik.com from 81.169.146.162 cause and error)
MIME-Version: 1.0
In-Reply-To: <e8301ce3-fe27-183d-2883-4d27604d981c@gmail.com>
References: <acab7c00-f7b2-9188-a374-96f1dad45029@gmail.com>
 <7e336645-7b47-2772-257a-a51b6849e4b5@gmail.com> <d6d3194c-68e5-1e2d-636d-2e7603916a4f@gmail.com>
 <e8301ce3-fe27-183d-2883-4d27604d981c@gmail.com>
Date: Sun, 13 Aug 2017 16:31:26 +0200
Message-ID: <CANUQDCjYcnpWr_LpFMzZPsmj1zsS4OuvbqW8EPMEzeegBS3G1A@mail.gmail.com>
To: Stanislav Malyshev <smalyshev@gmail.com>
Cc: Rowan Collins <rowan.collins@gmail.com>, PHP Internals <internals@lists.php.net>
Content-Type: multipart/alternative; boundary="94eb2c0b9150bd63300556a366e0"
Subject: Re: [PHP-DEV] Stop Exceptions capturing object references for trace arguments
From: me@kelunik.com (Niklas Keller)

--94eb2c0b9150bd63300556a366e0
Content-Type: text/plain; charset="UTF-8"

2017-08-12 17:54 GMT+02:00 Stanislav Malyshev <smalyshev@gmail.com>:

> Hi!
>
> > My only concern is if anyone is using this data for anything other than
> > debugging - e.g. if they were somehow extracting extra context about the
> > exception by traversing the backtrace to a particular point and grabbing
> > the arguments. That feels like a horrible hack to me, but that doesn't
> > mean someone isn't relying on it.
>
> Sure, that's why I propose to keep the option. If you use the horrible
> hack, I think it wouldn't be too much to ask to set an INI variable.
> Since it's rather local, it can be INI_ALL so you could do it in the
> script even.
>
> > An interesting side effect of an ini setting, though, is that it could
> > in turn be made to raise deprecation warnings at some point, whereas
> > just accessing the 'args' element of an array is pretty hard to detect.
> > So for instance, we could theoretically have: 7.3, add ini setting
> > defaulting to on; 7.4, change default to off and raise deprecation
> > notice if it's turned on; 8.0, remove feature, error if ini setting is
> > not set to on (ignore if set to off for smoother upgrades).
>
> Yep, sounds like a plan. Not sure if we need to drop it for 8.0 even,
> but we can vote on that. Anybody to make an RFC for this? :)
>

If it's explicitly needed, someone could still just call
`debug_backtrace()` and manually store the args in the exception
constructor, no?

Would the args still be available in the stringified version? Because
they're quite useful for debugging there.

Regards, Niklas

--94eb2c0b9150bd63300556a366e0--