Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:122170
MIME-Version: 1.0
References: <CAH5C8xXPNz1e2EARB5HpuiWpuRh=qRj8r1ZgpCH-=UFRJ_9O=Q@mail.gmail.com>
 <CAEKnhAEHuA-r+fwNVK7Mv8P43N4F8z+4QtXz1DnQawrtQyCw+w@mail.gmail.com>
In-Reply-To: <CAEKnhAEHuA-r+fwNVK7Mv8P43N4F8z+4QtXz1DnQawrtQyCw+w@mail.gmail.com>
Date: Wed, 17 Jan 2024 22:14:00 +0100
Message-ID: <CAH5C8xUR2Sky3Xf8p+NTMj_+RgQfkrMn0YUDd_=4K=-8qp8KEA@mail.gmail.com>
To: Jakub Zelenka <bukka@php.net>
Cc: PHP Internals List <internals@lists.php.net>
Content-Type: multipart/alternative; boundary="000000000000723b5a060f2aba52"
Subject: Re: [PHP-DEV] Dedicated StreamBucket class
From: kocsismate90@gmail.com (=?UTF-8?B?TcOhdMOpIEtvY3Npcw==?=)

--000000000000723b5a060f2aba52
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

Hi Jakub,

I just submitted feedback to the PR but will also mention it here as it's
> probably more an API thing. The problem that I see is that it combines tw=
o
> distinct things and create quite ugly self reference inside the proposed
> StreamBucket object. I would prefer we split it and introduce
> StreamBucketHandle opaque object that will completely replace the current
> use of bucket resource. The actual StreamBucket could be introduced later
> (ideally in PHP 9 as it's a sort of breaking change - change of class).
>

I agree with using a two step migration approach, and that's why I only
changed one part for now:
the containing object, leaving its resource property intact. However, as
far as I can tell,
there's no need to create a separate StreamBucketHandle object when
migrating the stream bucket resource to an
object, but we could inline it directly into the containing StreamBucket
object. I'm saying this because the
stream bucket resource property is only used by two functions:
stream_bucket_append() and stream_bucket_prepend(),
and I cannot imagine any other use-case where having a
StreamBucket::$bucket property would be useful (but correct
me if I'm wrong, I'm not the most experienced with stream buckets).


> I think it's more typing issue if someone passes this object for further
> processing and type hint stdClass. Possibly the pattern above could be us=
ed
> for copying but seems quite unlikely. Still I would see this as a BC brea=
k
> and it is not really related to resource to object migration.
>

For me, it seems like a subtle edge case which could be addressed by
explicitly mentioning the change in the UPGRADING file.
But I got your point, and I'm ok to submit an RFC.

Regards,
M=C3=A1t=C3=A9

--000000000000723b5a060f2aba52--