Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:116477
User-Agent: Cyrus-JMAP/3.5.0-alpha0-1371-g2296cc3491-fm-20211109.003-g2296cc34
Mime-Version: 1.0
Message-ID: <4c685dfa-2a81-4eed-af70-3e34b7de50f0@www.fastmail.com>
In-Reply-To: 
 <CAH5C8xUn9fD5=hkKQL+CMWSMXEugpX=HbSRcWQ3WTksCpXdt_A@mail.gmail.com>
References: 
 <CAH5C8xUn9fD5=hkKQL+CMWSMXEugpX=HbSRcWQ3WTksCpXdt_A@mail.gmail.com>
Date: Mon, 22 Nov 2021 11:52:10 -0600
To: "php internals" <internals@lists.php.net>
Content-Type: text/plain;charset=utf-8
Content-Transfer-Encoding: quoted-printable
Subject: Re: [PHP-DEV] [RFC] [Discussion] Readonly Classes
From: larry@garfieldtech.com ("Larry Garfield")

On Mon, Nov 22, 2021, at 10:13 AM, M=C3=A1t=C3=A9 Kocsis wrote:
> Hi Internals,
>
> I'd like to propose adding support for readonly classes:
> https://wiki.php.net/rfc/readonly_classes
>
> The implementation heavily builds on the semantics of the already acce=
pted
> readonly properties RFC (https://wiki.php.net/rfc/readonly_properties_=
v2),
> basically the only thing I had to implement in the PR are the followin=
g:
> - Any declared property of a readonly class is implicitly treated as
> readonly
> - Creation of dynamic properties is forbidden
>
> Readonly classes would make it easier to declare immutable(-alike) cla=
sses,
> especially if a class
> contains a lot of properties. As a side-effect, one could already opt =
out
> from dynamic properties
> completely, which won't really be possible for quite some time, even if
> https://wiki.php.net/rfc/deprecate_dynamic_properties passes.
>
> Regards:
> M=C3=A1t=C3=A9

If I'm following, then this RFC is about 90% syntactic sugar for putting=
 `readonly` on all properties, plus disabling dynamic properties.  That'=
s the long-and-short of it, yes?

Does disabling dynamic properties offer any other advantages as have bee=
n discussed otherwise recently?  (E.g., performance.)

Is there a way to opt a given property back OUT of being readonly, or if=
 you have a readonly class and need to add a single mutable property to =
it are you stuck adding `readonly` to all existing properties first?  (I=
'm not sure I'm suggesting having such a mechanism, mostly just clarifyi=
ng.)

It's interesting that this also provides a backdoor way to force all pro=
perties to be typed, as a side effect.  However... that also means you c=
annot mark a class readonly if any of its properties are callables, sinc=
e callables cannot be typed. (I guess you could use mixed and then docbl=
ock callable, which is a bit fugly but not a problem introduced by this =
RFC.)

--Larry Garfield