Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:123981
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable
Precedence: bulk
Mime-Version: 1.0 (1.0)
Subject: Re: [PHP-DEV] Overriding GMP objects
Date: Fri, 28 Jun 2024 09:43:26 +0900
Message-ID: <38C21842-1740-48A2-B2F1-0FFA57D542F4@sakiot.com>
References: <e665bb60-8697-40ec-9e8b-d83e0b8d5655@app.fastmail.com>
Cc: "Gina P. Banyard" <internals@gpb.moe>, internals@lists.php.net
In-Reply-To: <e665bb60-8697-40ec-9e8b-d83e0b8d5655@app.fastmail.com>
To: Rob Landers <rob@bottled.codes>
From: saki@sakiot.com (Saki Takamachi)

Hi Rob,

>> On Thu, Jun 27, 2024, at 06:07, Saki Takamachi wrote:
>>=20
>> I agree with Gina. Being able to change the class of a computed result of=
 an inherited child class causes several problems.
>>=20
>> The points raised in the `BCMath\Number` discussion can be summarized as f=
ollows.
>>=20
>> - If it is possible to perform calculations on parent Class and child Cla=
ss, what should the resulting class be?
>=20
> Ask the class what it wants to be?
>=20
>> - Multiplying the distance class and the speed class should give you the t=
ime class.
>=20
> That would depend on the library=E2=80=99s implementation.=20
>=20
>> - Similarly, multiplying two distance classes together should yield an ar=
ea class.
>=20
> See above.=20
>=20
>> - If only child classes have a property that should be specified in the c=
onstructor, how do you determine the value of this property in the result cl=
ass? Perhaps it is possible to determine the value of the property, but in t=
hat case the commutativity of the computation should be lost.
>=20
> This can be handled via the child class in static methods, per the library=
 specifications.
>=20
>>=20
>> These problems cannot be solved without significant complexity. And there=
 is a possibility that it cannot be resolved in the first place.
>=20
> I very much disagree; there is very little complications in the extension.=
 It=E2=80=99s actually quite simple:
>=20
> If one value is scalar and the other an instance of GMP, for binary ops, a=
sk the GMP class to perform the operation. (This is a one-linish change)
>=20
> If both are GMP instances, in a binary op, ask the left-most one to perfor=
m the op. The default GMP implementation will delegate to the right-side if t=
he right side isn=E2=80=99t a base-GMP instance, otherwise, business as usua=
l.
>=20
> And that=E2=80=99s pretty much it, other than defining the base-type ops. =
=46rom there, you can implement any units library you want, with whatever be=
havior makes sense =E2=80=94 so long as the base representation is a number.=

>=20
> If I remember the original operator overloading RFC, this is exactly what w=
as asked for by the people who voted no. In theory, it should pass if brough=
t as an RFC.=20
>=20
> =E2=80=94 Rob

I see.

All of what I wrote are problems that arise when php doesn't provide a way t=
o control them in userland.

Your proposal implicitly allows operator overloading in userland.

So your proposal essentially amounts to allowing operator overloading only f=
or GMP classes.

If I understand correctly, this discussion is not centered on GMP, but on th=
e topic of limited exposure of operator overloading functionality.

I believe this is something that requires some very careful discussion.

Regards,

Saki=