Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:122100
Date: Wed, 03 Jan 2024 14:45:55 +0000
To: Aleksander Machniak <alec@alec.pl>
Cc: internals@lists.php.net
Message-ID: <RlC3PYtRYWJ5sw6cA6PlI03E5lOd7dw43xzPF5nYuxnfpIpsQUlPA58_GtEvxbJf106xBrdYip7rkcfknhZ4bx30CXjgZnPnYC62CvOzsPE=@gpb.moe>
In-Reply-To: <648ce934-6203-4f17-8fbc-dc1f51c78e56@alec.pl>
References: <JmodrpwuVz43pzkm6Ok1P6SNx-ayT9ZJ1EvPazHh1Jqp7uT-RsqomzHc2-YYC2Qhq6DqJzWcltUMxo0xnAxpEWi8gccFPwJN7t43vkchbl8=@gpb.moe> <648ce934-6203-4f17-8fbc-dc1f51c78e56@alec.pl>
Feedback-ID: 96993444:user:proton
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable
Subject: Re: [PHP-DEV] [RFC] Add http_(get|clear)_last_request_headers() function
From: internals@gpb.moe ("Gina P. Banyard")

On Wednesday, 3 January 2024 at 14:34, Aleksander Machniak <alec@alec.pl> w=
rote:

> On 3.01.2024 14:41, Gina P. Banyard wrote:
>
> > Link: https://wiki.php.net/rfc/http-last-response-headers
>
>
> Wrong function name in the subject (should be "response" not "request")

Ah yes indeed, the RFC title is also incorrect.

> I don't think we need the clearing function. Do we?

Yes we do, because this is effectively global state whereas the variable is=
 created in the *local* scope.

> I don't like that this is HTTP specific feature while we have other
> protocol wrappers. Here's a different approach. Use stream context with
> extended context parameters feature. Something like:
>
> $context =3D stream_context_create();
>
> $file =3D file_get_contents('http://www.example.com/', false, $context);
>
> $headers =3D stream_context_get_params($context)['response_headers'];
>
> Or something like that. I don't know.
>
> While on this I found out that we already have stream_get_meta_data()
> and `wrapper_data` there. So, maybe we should/could make it more unified.
>
> Maybe it should be mentioned in the RFC.

The whole point of this RFC is to be able to remove the local variable, req=
uiring to drop to the stream layer, is the reason this got removed from the=
 8.1 mass deprecation RFC.
That RFC also shows how this can be achieved currently, but it is way more =
cumbersome.

You cannot use any of the stream related functions if the request fails unl=
ess one ignores errors, which requires parsing every single response to kno=
w if it is valid or not instead of just checking for false.

Moreover, this means one cannot use file_get_contents().

As such, I am not going to change anything in the RFC as my objective is to=
 get rid of the local variable creation in the long term.


Best regards,

Gina P. Banyard