Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:125971
Feedback-ID: ifab94697:Fastmail
Precedence: bulk
MIME-Version: 1.0
Date: Sat, 16 Nov 2024 10:23:49 +0100
To: =?UTF-8?Q?Marco_Aur=C3=A9lio_Deleu?= <deleugyn@gmail.com>
Cc: internals@lists.php.net
Message-ID: <98d66def-55cd-43f3-9d01-e2c58f5c0869@app.fastmail.com>
In-Reply-To: <BFEF1623-CB82-48DC-8F63-4A649A277222@gmail.com>
References: <b98dc200-02b7-468e-a96d-a38ce10656ed@app.fastmail.com>
 <BFEF1623-CB82-48DC-8F63-4A649A277222@gmail.com>
Subject: Re: [PHP-DEV] Optional interfaces
Content-Type: multipart/alternative;
 boundary=ea888a70899b4e9e8da59c2b7a61bb98
From: rob@bottled.codes ("Rob Landers")

--ea888a70899b4e9e8da59c2b7a61bb98
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable

On Sat, Nov 16, 2024, at 01:27, Marco Aur=C3=A9lio Deleu wrote:
>=20
>> On 15 Nov 2024, at 20:42, Rob Landers <rob@bottled.codes> wrote:
>> =EF=BB=BF
>> On Sat, Nov 16, 2024, at 00:34, Juris Evertovskis wrote:
>>> Hey all,
>>> =20
>>> If you try to implement an interface that does not exist, you get th=
e =E2=80=98Interface =E2=80=9C%s=E2=80=9D not found=E2=80=99 error. Usua=
lly that=E2=80=99s useful as it points to a code error.
>>> =20
>>> However, sometimes we want our code to be compatible with an optiona=
l library. For example, my Expression class might be compatible with som=
e ORM, but it would also usable without it or with another ORM. This can=
 already be accomplished, but the solutions seems very workaroundish: ht=
tps://stackoverflow.com/questions/57996429/implements-interface-only-if-=
exists
>>> =20
>>> Maybe the PHP itself should allow marking some interfaces as optiona=
l =E2=80=94 such that they are discarded from the list if the interface =
class does not exist? Here=E2=80=99s one way the syntax could look like:
>>> =20
>>>     class Expression implements MyInterface, ?OptionalInterface {}
>>> =20
>>>     interface EventDispatcherInterface extends ?PsrEventDispatcherIn=
terface {}
>>> =20
>>> BR,
>>> Juris
>>=20
>> Hey Juris,
>>=20
>> Is there some reason you cannot do this:
>>=20
>> https://3v4l.org/hF1vr
>>=20
>> =E2=80=94 Rob
>=20
> That seems to match what he shared on SO link?

Thanks Marco,

I don't visit stackoverflow because I loath "features locked behind inte=
rnet points". Thanks for letting me know what is behind the link!

>=20
> I think I opened this email with a strong negative feeling towards wha=
t would be optional interface, but to be honest a syntax sugar where imp=
lements ?MyInterface becomes something pretty close to what Rob shared d=
oesn=E2=80=99t sound bad, it=E2=80=99s just convenience. Unfortunately P=
HP is not keen on being open to those.
>=20
> Overall, I dig it.

I maintain a PHP SDK that needs to be compatible with Symfony and Larave=
l and any other framework. The shenanigans are all bound in one file tha=
t I very rarely touch. So, I do kinda like the idea of having optional i=
nterfaces, but I feel like it is rather niche. As an application develop=
er, I can't think of one time I've ever needed this.

However, the single biggest advantage comes from IDEs that can somehow s=
how that there is an optional interface on a type. I think that would be=
 the most useful bit to come from this, honestly. It also prevents IDEs =
or static analysis from giving a warning that there are multiple definit=
ions of an interface if you have the appropriate libraries installed.

It's also rather simple to implement, so I think it is worth pursuing, I=
MHO.

=E2=80=94 Rob
--ea888a70899b4e9e8da59c2b7a61bb98
Content-Type: text/html; charset=utf-8
Content-Transfer-Encoding: quoted-printable

<!DOCTYPE html><html><head><title></title><style type=3D"text/css">p.Mso=
Normal,p.MsoNoSpacing{margin:0}</style></head><body><div>On Sat, Nov 16,=
 2024, at 01:27, Marco Aur=C3=A9lio Deleu wrote:<br></div><blockquote ty=
pe=3D"cite" id=3D"qt" style=3D""><div dir=3D"ltr"><br></div><div dir=3D"=
ltr"><blockquote type=3D"cite"><div>On 15 Nov 2024, at 20:42, Rob Lander=
s &lt;rob@bottled.codes&gt; wrote:<br></div></blockquote></div><blockquo=
te type=3D"cite"><div dir=3D"ltr"><div>=EF=BB=BF<br></div><div>On Sat, N=
ov 16, 2024, at 00:34, Juris Evertovskis wrote:<br></div><blockquote typ=
e=3D"cite" id=3D"qt-qt" style=3D"overflow-wrap:break-word;"><div class=3D=
"qt-qt-WordSection1"><p class=3D"qt-qt-MsoNormal">Hey all,<br></p><p cla=
ss=3D"qt-qt-MsoNormal">&nbsp;<br></p><p class=3D"qt-qt-MsoNormal">If you=
 try to implement an interface that does not exist, you get the =E2=80=98=
Interface =E2=80=9C%s=E2=80=9D not found=E2=80=99 error. Usually that=E2=
=80=99s useful as it points to a code error.<br></p><p class=3D"qt-qt-Ms=
oNormal">&nbsp;<br></p><p class=3D"qt-qt-MsoNormal">However, sometimes w=
e want our code to be compatible with an optional library. For example, =
my Expression class might be compatible with some ORM, but it would also=
 usable without it or with another ORM. This can already be accomplished=
, but the solutions seems very workaroundish: <a href=3D"https://stackov=
erflow.com/questions/57996429/implements-interface-only-if-exists">https=
://stackoverflow.com/questions/57996429/implements-interface-only-if-exi=
sts</a><br></p><p class=3D"qt-qt-MsoNormal">&nbsp;<br></p><p class=3D"qt=
-qt-MsoNormal">Maybe the PHP itself should allow marking some interfaces=
 as optional =E2=80=94 such that they are discarded from the list if the=
 interface class does not exist? Here=E2=80=99s one way the syntax could=
 look like:<br></p><p class=3D"qt-qt-MsoNormal">&nbsp;<br></p><p class=3D=
"qt-qt-MsoNormal">&nbsp;&nbsp;&nbsp; class Expression implements MyInter=
face, ?OptionalInterface {}<br></p><p class=3D"qt-qt-MsoNormal">&nbsp;<b=
r></p><p class=3D"qt-qt-MsoNormal">&nbsp;&nbsp;&nbsp; interface EventDis=
patcherInterface extends ?PsrEventDispatcherInterface {}<br></p><p class=
=3D"qt-qt-MsoNormal">&nbsp;<br></p><p class=3D"qt-qt-MsoNormal">BR,<br><=
/p><p class=3D"qt-qt-MsoNormal">Juris<br></p></div></blockquote><div><br=
></div><div>Hey Juris,<br></div><div><br></div><div>Is there some reason=
 you cannot do this:<br></div><div><br></div><div><a href=3D"https://3v4=
l.org/hF1vr">https://3v4l.org/hF1vr</a><br></div><div><br></div><div id=3D=
"qt-sig121229152">=E2=80=94 Rob<br></div></div></blockquote><div><br></d=
iv><div>That seems to match what he shared on SO link?<br></div></blockq=
uote><div><br></div><div><div>Thanks Marco,<br></div><div><br></div><div=
>I don't visit stackoverflow because I loath "features locked behind int=
ernet points". Thanks for letting me know what is behind the link!<br></=
div></div><div><br></div><blockquote type=3D"cite" id=3D"qt" style=3D"">=
<div><br></div><div>I think I opened this email with a strong negative f=
eeling towards what would be optional interface, but to be honest a synt=
ax sugar where implements ?MyInterface becomes something pretty close to=
 what Rob shared doesn=E2=80=99t sound bad, it=E2=80=99s just convenienc=
e. Unfortunately PHP is not keen on being open to those.<br></div><div><=
br></div><div>Overall, I dig it.<br></div></blockquote><div><br></div><d=
iv>I maintain a PHP SDK that needs to be compatible with Symfony and Lar=
avel and any other framework. The shenanigans are all bound in one file =
that I very rarely touch. So, I do kinda like the idea of having optiona=
l interfaces, but I feel like it is rather niche. As an application deve=
loper, I can't think of one time I've ever needed this.<br></div><div><b=
r></div><div>However, the single biggest advantage comes from IDEs that =
can somehow show that there is an optional interface on a type. I think =
that would be the most useful bit to come from this, honestly. It also p=
revents IDEs or static analysis from giving a warning that there are mul=
tiple definitions of an interface if you have the appropriate libraries =
installed.<br></div><div><br></div><div>It's also rather simple to imple=
ment, so I think it is worth pursuing, IMHO.</div><div><br></div><div id=
=3D"sig121229152">=E2=80=94 Rob<br></div></body></html>
--ea888a70899b4e9e8da59c2b7a61bb98--