Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:128092
Feedback-ID: ifab94697:Fastmail
Precedence: bulk
MIME-Version: 1.0
Date: Thu, 17 Jul 2025 08:52:34 +0200
To: "Nicolas Grekas" <nicolas.grekas+php@gmail.com>
Cc: "Dmitry Derepko" <xepozzd@gmail.com>, internals@lists.php.net
Message-ID: <21a13437-9104-4950-a2dd-13b9cf1d66cf@app.fastmail.com>
In-Reply-To: 
 <CAOWwgpk2YqoDaC4O9wqCSJxRcsZKT=R0B6XLPA1PboFo=MMMkA@mail.gmail.com>
References: <d603f372-5f58-4308-aba1-151c35f3aa94@app.fastmail.com>
 <0D6532F3-6E95-48B9-B394-E9CC1EC00B56@gmail.com>
 <d4da4b69-7f13-4e8f-a043-d610d178479b@app.fastmail.com>
 <CAOWwgpk2YqoDaC4O9wqCSJxRcsZKT=R0B6XLPA1PboFo=MMMkA@mail.gmail.com>
Subject: Re: [PHP-DEV] RFC: Records
Content-Type: multipart/alternative;
 boundary=cc4d1520f5664594a79871409c308489
From: rob@bottled.codes ("Rob Landers")

--cc4d1520f5664594a79871409c308489
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable

On Tue, Jul 15, 2025, at 19:05, Nicolas Grekas wrote:
> Not sure it's really a contribution to the discussion but in case you =
missed it, you can make public properties "public(set)":
>=20
> final readonly class Response {
>     public function __construct(
>         public public(set) int $statusCode,
>         public public(set) string $reasonPhrase,
>         // ...
>     ) {
>         if($this->statusCode >=3D 600) {
>             throw new LogicException();
>         }
>     }
> }
>=20
> Then your example works.
> That's ugly, but that doesn't have to stay that way.
> And I'm now wondering why is that not the default?
>=20
> Nicolas

It "works" but it still results in an invalid state (statusCode >=3D 600=
 is never thrown). Previously, this kind of invalid object would only be=
 possible via reflection and skipping the constructor. Now, we have a co=
py, without a copy-constructor to perform any validation or error checki=
ng before the copy completes; so basically invalid objects are incredibl=
y easy/accidental to be made, unless everyone starts changing all valida=
tion to be done in getters/setters (which is not possible on readonly ob=
jects) or by keeping things protected(set) and ensuring that only specif=
ic functions set them... and probably make every readonly class final.

=E2=80=94 Rob
--cc4d1520f5664594a79871409c308489
Content-Type: text/html; charset=utf-8
Content-Transfer-Encoding: quoted-printable

<!DOCTYPE html><html><head><title></title></head><body><div>On Tue, Jul =
15, 2025, at 19:05, Nicolas Grekas wrote:<br></div><blockquote type=3D"c=
ite" id=3D"qt" style=3D""><div dir=3D"ltr"><div class=3D"qt-gmail_quote =
qt-gmail_quote_container"><div>Not sure it's really a contribution to th=
e discussion but in case you missed it, you can make public properties "=
public(set)":</div><div><br></div><div><div>final readonly class Respons=
e {</div><div>&nbsp; &nbsp; public function __construct(</div><div>&nbsp=
; &nbsp; &nbsp; &nbsp; public public(set) int $statusCode,</div><div>&nb=
sp; &nbsp; &nbsp; &nbsp; public public(set) string $reasonPhrase,</div><=
div>&nbsp; &nbsp; &nbsp; &nbsp; // ...</div><div>&nbsp; &nbsp; ) {</div>=
<div>&nbsp; &nbsp; &nbsp; &nbsp; if($this-&gt;statusCode &gt;=3D 600) {<=
/div><div>&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; throw new LogicExcep=
tion();</div><div>&nbsp; &nbsp; &nbsp; &nbsp; }</div><div>&nbsp; &nbsp; =
}</div><div>}</div></div><div><br></div><div>Then your example works.</d=
iv><div>That's ugly, but that doesn't have to stay that way.</div><div>A=
nd I'm now wondering why is that not the default?</div></div><div class=3D=
"qt-gmail_quote qt-gmail_quote_container"><br></div><div class=3D"qt-gma=
il_quote qt-gmail_quote_container">Nicolas</div></div></blockquote><div>=
<br></div><div>It "works" but it still results in an invalid state (stat=
usCode &gt;=3D 600 is never thrown). Previously, this kind of invalid ob=
ject would only be possible via reflection and skipping the constructor.=
 Now, we have a copy, without a copy-constructor to perform any validati=
on or error checking before the copy completes; so basically invalid obj=
ects are incredibly easy/accidental to be made, unless everyone starts c=
hanging all validation to be done in getters/setters (which is not possi=
ble on readonly objects) or by keeping things protected(set) and ensurin=
g that only specific functions set them... and probably make every reado=
nly class final.<br></div><div><br></div><div id=3D"sig121229152">=E2=80=
=94 Rob</div></body></html>
--cc4d1520f5664594a79871409c308489--