Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:113214
Content-Type: multipart/alternative;
	boundary="Apple-Mail=_05820433-D6B3-499A-A8B4-4D8D09D4DD04"
Mime-Version: 1.0 (Mac OS X Mail 14.0 \(3654.60.0.2.21\))
Date: Mon, 22 Feb 2021 11:14:15 +0100
References: <52E47CD0-11F0-45DF-9387-DFB5DFDC481A@cschneid.com>
 <CAESVnVoMmQZCUaUTkYDzGHPSoS9meJyuTm045TO=O20-PLmqLA@mail.gmail.com>
To: PHP internals <internals@lists.php.net>
In-Reply-To: <CAESVnVoMmQZCUaUTkYDzGHPSoS9meJyuTm045TO=O20-PLmqLA@mail.gmail.com>
Message-ID: <A093EB87-93C5-4C61-9C28-E28FAC45EE8F@cschneid.com>
Subject: Re: [PHP-DEV] Small annoyances of PHP: Simple fixes would make tests
 fail
From: cschneid@cschneid.com (Christian Schneider)

--Apple-Mail=_05820433-D6B3-499A-A8B4-4D8D09D4DD04
Content-Transfer-Encoding: quoted-printable
Content-Type: text/plain;
	charset=us-ascii

Am 19.02.2021 um 16:21 schrieb Sara Golemon <pollita@php.net>:
> On Fri, Feb 19, 2021 at 8:04 AM Christian Schneider =
<cschneid@cschneid.com <mailto:cschneid@cschneid.com>> wrote:
> Stack traces:
> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
>=20
> What version are you testing with. I know Niki killed some unnecessary =
frames in 8.0's stacktrace output. (A whole repro script is more useful =
than a snippet wherein $e comes into being seemingly from nowhere).

I was testing with PHP 8.0.3 but it is the same with master.
You can see the difference at https://3v4l.org/0Pgfo =
<https://3v4l.org/0Pgfo> where the first line
	try { $a[(object)null] =3D 1; } catch (Throwable $e) { =
print_r($e->getTrace()); }
does not create a stack trace. That's what I'd like to change.

What do you think? Should I create a PR for this?
It does break a lot of tests in PHP. Probably less so for other projects =
but still...

> Indentation in var_export:
> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
> The indentation for
>         var_export((object)[(object)[]]);
> is off by one for nested structures:
>         (object) array(
>            '0' =3D>                       # <-- extra space before =
index
>           (object) array(
> and the fix would be to change
>         buffer_append_spaces(buf, level + 2)
> to
>         buffer_append_spaces(buf, level + 1)
> in ext/standard/var.c
>=20
> This breaks about 60 PHP tests relying on the var_export format.
>=20
>=20
> I think shifting the indentation in var_export() could probably be =
done without a major BC break.

An example script is at https://3v4l.org/V8r5k <https://3v4l.org/V8r5k>
It does break tests though. And my guess it that is would also break =
tests in quite some projects.

What version would you aim this for? 8.1? 9.0?

> Trimming trailing whitespace would also resolve one of my long-time =
annoyances with the function.
> Feel free to propose that as a PR and/or an RFC.

I see your point as it means there have to be trailing white-spaces in =
tests which can be confusing when editing/creating tests.
I can have a look at how hard this would be to fix...

Side-note: Maybe we should recommend a function to be used for tests. At =
my work we used to use var_export(), some of our newer tests are using =
json_encode() but neither is really guaranteed to be stable, right?

- Chris


--Apple-Mail=_05820433-D6B3-499A-A8B4-4D8D09D4DD04--