Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:118717
MIME-Version: 1.0
In-Reply-To: <CANS-=pfikfZAbRtugObswwGFqQ3VAje7W7t3J6yih7XTDytevA@mail.gmail.com>
References: <CAJpLVh2y1SKiuNzbqNC=RrNUjKaLGMNfy4VjE-92q8LgWOqTsg@mail.gmail.com>
 <CAC_PL4wGQCWE8S_eO2PqskSuL=P0xEzUou0F=AfThwStMPWrdw@mail.gmail.com>
 <CAJpLVh0jRgGsC83D80r-JzmnTjhAjr_K9EaqxYorHirNrTLJ6Q@mail.gmail.com> <CANS-=pfikfZAbRtugObswwGFqQ3VAje7W7t3J6yih7XTDytevA@mail.gmail.com>
Date: Fri, 30 Sep 2022 16:53:05 +0200
Message-ID: <CAJpLVh2+jS8+z_B_U_yprHOihvW4xTKkm-3QuYhfcB37h2v3TA@mail.gmail.com>
To: Hamza Ahmad <office.hamzaahmad@gmail.com>, internals@lists.php.net
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Subject: Re: [PHP-DEV] Re: Alias for `int|float`
From: olleharstedt@gmail.com (=?UTF-8?Q?Olle_H=C3=A4rstedt?=)

2022-09-30 12:15 GMT+02:00, Lynn <kjarli@gmail.com>:
> On Fri, Sep 30, 2022 at 11:04 AM Olle H=C3=A4rstedt <olleharstedt@gmail.c=
om>
> wrote:
>
>> 2022-09-29 5:08 GMT+02:00, Hamza Ahmad <office.hamzaahmad@gmail.com>:
>> > Hi Olle,
>> >
>> > I appreciate your idea of introducing a similar concept of typedef.
>> > What if you write an RFC explaining that concept. I can join you
>> > however in co-authoring this request.
>> >
>> > Seriously, I have had issues with writing such type again and again.
>> > If you look at mixed type, it is indeed an alias of some union types.
>> > Adding such an option to userland would lead to introduction of
>> > various type hints. Plus, it will help devs write short hand aliases
>> > for their intersection types.
>>
>> Note that my suggestion here is NOT about a project-global type alias,
>> but an extension of the existing file-only alias, the use-statement.
>> This can also explain the lack of reaction. :) Internal devs might
>> already have consensus on the right way to proceed here, even if
>> there's no RFC yet.
>>
>> Olle
>>
>> --
>> PHP Internals - PHP Runtime Development Mailing List
>> To unsubscribe, visit: https://www.php.net/unsub.php
>>
>>
> I don't know if file based type aliases make sense, but just spitballing
> some ideas here.
> ```
> // core php or perhaps some libraries
> typedef number =3D int|float;
> typedef number<number $min> =3D TypeDef::extend('number', static fn (numb=
er
> $value, number $min) =3D> $value >=3D $min);
> typedef number<?number $min, number $max> =3D TypeDef::extend('number',
> static fn (number $value, ?number $min, number $max) =3D> ($min =3D=3D=3D=
 null ||
> $value >=3D $min) && $value <=3D $max);
> typedef even =3D TypeDef::extend('number<?number $min, number $max>', sta=
tic
> fn (number $value) =3D> $value % 2 =3D=3D=3D 0);
> typedef positive =3D number<0>;
> typedef negative =3D number<max: 0>;
> typedef fraction =3D float<0, 1>;
> typedef string<int $min> =3D TypDef::extend('string', static fn (string
> $value, int $min) =3D> strlen($value) >=3D $min);
> typedef string<?int $min, int $max> =3D TypeDef::extend('string', static =
fn
> (string $value, ?int $min, int $max) =3D> length_is_between($value, $min,
> $max));
> typedef regex<string $expression> =3D TypeDef::extend('string', static fn
> (string $value, string $regex) =3D> preg_match($regex, $value));
>
> namespace  App\Domain;
>
> typedef NonEmptyString =3D string<1>; // minimum length of 1
> typedef LimitedString =3D string<null, 32>; // no minimum length, max len=
gth
> of 32
> // or typedef LimitedString =3D string<max: 32>; as alternative notation =
with
> named parameters
> typedef SomeToken =3D regex<'/a-z0-9{12}/i'>;
>
> // only allows even numbers ranging from -100 to 100
> function someFunction(even<-100, 100> $number) {}
>
> namespace Elsewhere;
>
> use  App\Domain\NonEmptyString;
> use  App\Domain\LimitedString;
> use  App\Domain\SomeToken;
>
> someFunction(102); // fails `$max` check from `number<min, max>`
> someFunction(-99); // fails `% 2` check from `even`
>
> ```
>
> I tried to use some consistency for the examples, by no means this is wha=
t
> I think it must look like, just making sure the intent of the examples is
> clear. This would be nothing more than syntactic sugar to what I'd
> otherwise write in custom objects to validate types and values and ensure
> type safety. This also doesn't take into account how operators should
> behave or how autoloading could/should behave.

Impressive repertoire of ideas, but probably out of scope from what I
was thinking of. :) Some call this "refinement types":
https://en.wikipedia.org/wiki/Refinement_type

Olle