Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:116976
To: php internals <internals@lists.php.net>
Thread-Topic: [PHP-DEV] Re: Adding `final class Deque` to PHP
Thread-Index: AQHXraUxaK1dZ0niuECOJlLUpHNcpax/lZKygAAgIQCAAXQbEIAAIukAgACdooY=
Date: Thu, 3 Feb 2022 03:14:05 +0000
Message-ID:
 <DM6PR14MB4155903D88D662A3BD756545F9289@DM6PR14MB4155.namprd14.prod.outlook.com>
References:
 <DM6PR07MB66185E23B803C5E5BF858B30F9DF9@DM6PR07MB6618.namprd07.prod.outlook.com>
 <DM6PR14MB41556131E6A9D7812854AA75F9269@DM6PR14MB4155.namprd14.prod.outlook.com>
 <1c9e79d5-fd79-45a8-9c77-4228d770062d@www.fastmail.com>
 <DM6PR14MB4155E8620E5907BDA2320B7EF9279@DM6PR14MB4155.namprd14.prod.outlook.com>
 <3ae1a3f5-b4ea-40a0-b03f-3fa69ed6b9d2@www.fastmail.com>
In-Reply-To: <3ae1a3f5-b4ea-40a0-b03f-3fa69ed6b9d2@www.fastmail.com>
Accept-Language: en-CA, en-US
Content-Language: en-CA
suggested_attachment_session_id: 580115f0-d8b0-9ef1-7f14-61087e6bce44
x-ms-exchange-messagesentrepresentingtype: 1
x-tmn:
 [BtM7mNDyW9ySQYh3A05GJXKXFkP/R/i9usaylYuZwiFE6B3IRlq4W0RwHp46mItKlk5F4dx7380=]
x-ms-publictraffictype: Email
x-ms-office365-filtering-correlation-id: ac8e4592-297d-439c-fcd6-08d9e6c33c6b
x-ms-traffictypediagnostic: MN2PR14MB2845:EE_
x-microsoft-antispam: BCL:0;
x-microsoft-antispam-message-info:
 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
x-ms-exchange-antispam-messagedata-chunkcount: 1
x-ms-exchange-antispam-messagedata-0:
 =?iso-8859-1?Q?a0xip/cjlNwOQxthINXt4GBzV9qUtCUQe4/iXW54QNwVwT7RP6j0RrRu9m?=
 =?iso-8859-1?Q?k7i7ZPiG5z/e/y0reOw4esCfY674gAPP8pIkr2i9SXFQReNudyl8F1Rv8j?=
 =?iso-8859-1?Q?BvhNP0lpqGqX54uQzolE1E/sT+IZG2qLedfmXJYP2yrN5P9Kw5jZLpaLtN?=
 =?iso-8859-1?Q?8rQHTF1FKvv3PvjY5KIo7eQwG1VSrWtatBrHSkociauO0P5G7oF74syoeG?=
 =?iso-8859-1?Q?uMzyQcIl1Vfpj7sTHEoyOkmzXIlUqoctYSWgFX9QN6HG8cadiBzfmiAv70?=
 =?iso-8859-1?Q?7cOsH6jx0QlfFmH9yOlrglIcE6zwKYzEafcDY23M8609/pHNW78QyCt9RN?=
 =?iso-8859-1?Q?irLtbyTO7fp36BaGFttSHJHnRg82mD6dRV/vZB4LIac+W/arrxNih0VqEh?=
 =?iso-8859-1?Q?HAFGVTx8JRwVnyegWkp19VLo22sW73FU/5Cvj/X+Oos7Hid8lM/50CR0+d?=
 =?iso-8859-1?Q?13vfE61Wd65UALiW9GBmvB1rmIebIVIBmkf5kYpNyZY3hncLRTaAMgWNNt?=
 =?iso-8859-1?Q?cAP/X9grZq1BZexH5MHgz7sKkn9n/LcasOdvYHaR8SmdUp0LADeWKnBFbB?=
 =?iso-8859-1?Q?SxFXbRw0H8H+IEwk5Ul/aa7PjglpVuboZXNuo3wVa8mKfuQO5dH0nIGkl7?=
 =?iso-8859-1?Q?OTgwD0l5WS7xPOssrbA9t6ilHHoRS/0DveNpoI2wsYTAEsWVLTkl9hjoI7?=
 =?iso-8859-1?Q?vx6y1iRVJyqBJ3jdWt8IziQqnYj6XTi/jWDrd2xQ8/MZFXxaJu92gWgUGZ?=
 =?iso-8859-1?Q?q9MhPn9ixwplFK6+KHyHuSX63DEy0duSwUnzkxGo4+TjCi6h5+5Nww9JvI?=
 =?iso-8859-1?Q?UCOzOIO7N8bUqI2O6MwBIYGO/0fdPV3CZmfU70jqUnfEj6V5R+jJoXff4C?=
 =?iso-8859-1?Q?BS1J5+VuQMiKvHqWOkpGePfPwMi50PazALqSOwXV+tba+IEd/3LNbzymcT?=
 =?iso-8859-1?Q?hJjUbxMte7pyqE1BpYyHk4nIGyOTgcZQNs45EpXsWHSVy6mxXxsXNS3EiT?=
 =?iso-8859-1?Q?ylPAFLgY8/gvfWDLNIPB7u5FuDyLIPUzhMNu67wJAVISbZ9wymPeZbwPp9?=
 =?iso-8859-1?Q?dA=3D=3D?=
Content-Type: text/plain; charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
MIME-Version: 1.0
X-OriginatorOrg: sct-15-20-4755-11-msonline-outlook-cd57b.templateTenant
X-MS-Exchange-CrossTenant-AuthAs: Internal
X-MS-Exchange-CrossTenant-AuthSource: CH2PR14MB4151.namprd14.prod.outlook.com
X-MS-Exchange-CrossTenant-RMS-PersistedConsumerOrg: 00000000-0000-0000-0000-000000000000
X-MS-Exchange-CrossTenant-Network-Message-Id: ac8e4592-297d-439c-fcd6-08d9e6c33c6b
X-MS-Exchange-CrossTenant-originalarrivaltime: 03 Feb 2022 03:14:05.5888
 (UTC)
X-MS-Exchange-CrossTenant-fromentityheader: Hosted
X-MS-Exchange-CrossTenant-id: 84df9e7f-e9f6-40af-b435-aaaaaaaaaaaa
X-MS-Exchange-CrossTenant-rms-persistedconsumerorg: 00000000-0000-0000-0000-000000000000
X-MS-Exchange-Transport-CrossTenantHeadersStamped: MN2PR14MB2845
Subject: Re: [PHP-DEV] Re: Adding `final class Deque` to PHP
From: tysonandre775@hotmail.com (tyson andre)

Hi Larry Garfield,=0A=
=0A=
> >> Returning $this would resolve that, I think.=A0 (Making it return a ne=
w, immutable copy of the Deque would be even nicer, but I realize that's pr=
obably not an argument I'm going to win at this point on this RFC.)=0A=
> >=0A=
> > Technically, you still can have single-expression usages in=0A=
> > readable/unreadable ways=0A=
> >=0A=
> > - `[$deque->shift('value'), $deque][1]`, or=0A=
> > - `($deque->shift('value') ?: $deque)`, or=0A=
> > - `my_userland_helper_shift_and_return($deque, 'value')`=0A=
>=0A=
> ... Those are all grotesque.=0A=
=0A=
True, I wasn't familiar with that event and whether your goal was to write =
as few lines as possible or as quickly as possible vs in a readable way to =
people familiar with php.=0A=
That clears that up.=0A=
=0A=
> > My personal preference is against making this fluent.=0A=
> > I'd rather expose an efficient datastructure that's consistent with the=
=0A=
> > rest of PHP's functionality to the extent possible,=0A=
> > which userland can use to write their own fluent/non-fluent classes.=0A=
> > There's drawbacks to returning `$this`, including:=0A=
> >=0A=
> > 1. Inconsistency with existing APIs making remembering what does what=
=0A=
> > harder. Barely anything in php that I remember returns $this.=0A=
> >=0A=
> >=A0=A0=A0 https://www.php.net/manual/en/arrayobject.append.php returns v=
oid.=0A=
> >=0A=
> >=A0=A0=A0 https://www.php.net/manual/en/function.array-push returns an i=
nt.=0A=
>=0A=
> So it's already inconsistent, and frankly neither of those returns is use=
ful.=A0 If the existing convention is "inconsistently unhelpful", then let'=
s go ahead and make it helpful since "it's lame but at least it's consisten=
t" is not an argument here.=A0 (Sometimes that is a valid argument, but not=
 in this case.)=0A=
=0A=
I won't know if anyone is strongly against fluent interfaces belonging in c=
ore, or strongly against adopting fluent interfaces in an ad-hoc way that w=
ould make it unclear if the vote was for/against the overall functionality =
or the new design choice.=0A=
=0A=
I'd rather not do this without a policy RFC such as the one that passed for=
 namespaces https://wiki.php.net/rfc/namespaces_in_bundled_extensions=0A=
=0A=
If you expect this to be widely considered a better design php should adopt=
, please create an RFC after the vote finishes (if it passes) before the fe=
ature freeze outlining which methods of `Collection\Deque` would change and=
 how.=0A=
=0A=
> > 2. Inconsistency with possible new datastructures/methods=0A=
> >=0A=
> >=A0=A0=A0 If a `Map`/`Set` function were to be added, then methods for=
=0A=
> > add/remove would return booleans (or the old value), not $this=0A=
>=0A=
> Removal methods need to return the value being removed, sure.=A0 That mak=
es sense across the board.=A0 But I don't see why Map::add($k, $v) or Set::=
add($v) shouldn't also return $this.=A0 I would make the exact same ask the=
re, for the exact same reason: To aid functional code, which (when the synt=
ax is conducive to it) can be more readable in many cases than procedural a=
pproaches.=0A=
>=0A=
> > 3. Slight additional performance overhead for functionality I assume=0A=
> > will be used relatively infrequently=0A=
>=0A=
> I would assume the opposite, at least in my own code.=A0 I'm writing a lo=
t of recursive or reduction-based routines these days, so if I had reason t=
o use a queue/stack in them in the first place, they'd almost certainly get=
 used in this fashion.=0A=
>=0A=
> >=A0=A0=A0 (php has to increment reference counts and opcache can't elimi=
nate=0A=
> > the opcode to decrease reference counts and possibly free the return=0A=
> > value of `$deque->shift()` with the return type info being an object)=
=0A=
>=0A=
> I cannot speak to this one; how much of a difference is it in practice?=
=A0=0A=
=0A=
Adding `RETURN_OBJ_COPY(Z_OBJ_P(ZEND_THIS));` and a different method with t=
he `: Deque` return type:=0A=
Around 6-8% for large arrays where push is the most frequent method call an=
d the array reads are taken out since they'd be the same?=0A=
(and similar for shift, I'd expect)=0A=
(Intel CPU security mitigations/power management and so on have made benchm=
arking less convenient, I ran the functions twice to see if it was consiste=
nt)=0A=
=0A=
I'd still object to this for other stated reasons even if you think 6-8% is=
 worth it for your use cases for this or future additions to php in general=
.=0A=
=0A=
```=0A=
Results for php 8.2.0-dev debug=3Dfalse with opcache enabled=3Dtrue=0A=
=0A=
Appending to Collections\Deque []=3D   : n=3D       4 iterations=3D10000000=
, create+destroy time=3D1.204 result=3D0=0A=
Appending to Collections\Deque push  : n=3D       4 iterations=3D10000000, =
create+destroy time=3D1.551 result=3D0=0A=
Appending to Collections\Deque fluent: n=3D       4 iterations=3D10000000, =
create+destroy time=3D1.640 result=3D0=0A=
Appending to Collections\Deque []=3D   : n=3D       4 iterations=3D10000000=
, create+destroy time=3D1.213 result=3D0=0A=
Appending to Collections\Deque push  : n=3D       4 iterations=3D10000000, =
create+destroy time=3D1.549 result=3D0=0A=
Appending to Collections\Deque fluent: n=3D       4 iterations=3D10000000, =
create+destroy time=3D1.636 result=3D0=0A=
=0A=
Appending to Collections\Deque []=3D   : n=3D       8 iterations=3D 5000000=
, create+destroy time=3D0.923 result=3D0=0A=
Appending to Collections\Deque push  : n=3D       8 iterations=3D 5000000, =
create+destroy time=3D1.263 result=3D0=0A=
Appending to Collections\Deque fluent: n=3D       8 iterations=3D 5000000, =
create+destroy time=3D1.335 result=3D0=0A=
Appending to Collections\Deque []=3D   : n=3D       8 iterations=3D 5000000=
, create+destroy time=3D0.921 result=3D0=0A=
Appending to Collections\Deque push  : n=3D       8 iterations=3D 5000000, =
create+destroy time=3D1.261 result=3D0=0A=
Appending to Collections\Deque fluent: n=3D       8 iterations=3D 5000000, =
create+destroy time=3D1.335 result=3D0=0A=
=0A=
Appending to Collections\Deque []=3D   : n=3D    1024 iterations=3D  100000=
, create+destroy time=3D1.246 result=3D0=0A=
Appending to Collections\Deque push  : n=3D    1024 iterations=3D  100000, =
create+destroy time=3D1.972 result=3D0=0A=
Appending to Collections\Deque fluent: n=3D    1024 iterations=3D  100000, =
create+destroy time=3D2.150 result=3D0=0A=
Appending to Collections\Deque []=3D   : n=3D    1024 iterations=3D  100000=
, create+destroy time=3D1.251 result=3D0=0A=
Appending to Collections\Deque push  : n=3D    1024 iterations=3D  100000, =
create+destroy time=3D1.969 result=3D0=0A=
Appending to Collections\Deque fluent: n=3D    1024 iterations=3D  100000, =
create+destroy time=3D2.146 result=3D0=0A=
=0A=
Appending to Collections\Deque []=3D   : n=3D 1048576 iterations=3D     100=
, create+destroy time=3D2.450 result=3D0=0A=
Appending to Collections\Deque push  : n=3D 1048576 iterations=3D     100, =
create+destroy time=3D3.172 result=3D0=0A=
Appending to Collections\Deque fluent: n=3D 1048576 iterations=3D     100, =
create+destroy time=3D3.371 result=3D0=0A=
Appending to Collections\Deque []=3D   : n=3D 1048576 iterations=3D     100=
, create+destroy time=3D2.455 result=3D0=0A=
Appending to Collections\Deque push  : n=3D 1048576 iterations=3D     100, =
create+destroy time=3D3.183 result=3D0=0A=
Appending to Collections\Deque fluent: n=3D 1048576 iterations=3D     100, =
create+destroy time=3D3.376 result=3D0=0A=
=0A=
```=0A=
=0A=
=0A=
> > 4.=A0 Returning $this makes code easier to write at some cost to=0A=
> > readability - Developers new to php or using `Collections\Deque` for=0A=
> > the first time would not immediately know what the code they're reading=
=0A=
> > is doing.=0A=
> >=A0=A0=A0 (less of a problem with a good IDE, typechecker, and a typed=
=0A=
> > codebase, but this isn't universal)=0A=
> >=A0=A0=A0 Having it return void, `return $deque->push()` would be less c=
ommon=0A=
> > and this would force the meaning to be clear.=0A=
>=0A=
> I disagree that it's less readable.=A0 Compare it with the alternatives y=
ou showed at the start of your reply.=A0 None of those would qualify as "re=
adable", IMO.=0A=
=0A=
True, I wasn't sure if your goal was readability (for php developers deeply=
 familiar with libraries) or to write as few lines as possible.=0A=
Suggesting using `function` and adding several lines seemed like something =
you were trying to avoid.=0A=
=0A=
=0A=
> >=A0=A0=A0 Developers might have several guesses/assumptions based on the=
ir=0A=
> > experience with other methods in php/elsewhere=0A=
> >=0A=
> >=A0=A0=A0 - It returns the new count (JavaScript Array.push, array_push)=
=0A=
> >=A0=A0=A0 - It returns $this (Ruby)=0A=
> >=A0=A0=A0 - It returns a lazy copy, like you'd wanted, not modifying the=
=0A=
> > original=0A=
> >=A0=A0=A0 - It's returning void and the code in question is shorthand fo=
r=0A=
> > `return null`.=0A=
> >=A0=A0=A0=A0=A0 (Python, C++=0A=
> > https://www.cplusplus.com/reference/vector/vector/push_back/ ,=0A=
> > offsetSet and spl push()/shift() methods)=0A=
>=0A=
> Once again, the lack of a consistent pattern means we don't need to follo=
w the pattern.=A0 As with the first point, "it's silly but at least it's co=
nsistent" is a valid argument in many cases, and we've changed the language=
 accordingly before.=A0 (Ternary associativity order comes to mind.)=A0 Tha=
t's not the case here.=A0 There is no broad-based ingrained training that p=
eople have that would lead them to expect a given return, so we can and sho=
uld go for whichever one is most powerful/flexible/conducive to good code.=
=A0 IMO, that's returning $this.=A0=0A=
=0A=
I continue to disagree for my previously stated reasons.=0A=
=0A=
> (Again, since a purely functional version is apparently off the table.)=
=0A=
=0A=
I'd be in favor of immutable datastructures in core, but I don't think I'd =
have much success in an RFC.=0A=
I've only heard from a few voters that are passionately for them, I'm not s=
ure if that translates to widespread interest.=0A=
=0A=
Though interestingly (and unrelated to this RFC https://wiki.php.net/rfc/de=
que ), they do seem possible to implement with better asymptotic performanc=
e than I'd have expected.=0A=
Google searching mentioned Relaxed Radix Binary Trees, an immutable datastr=
ucture surprisingly claiming performance near arrays in most operations.=0A=
https://hypirion.com/musings/thesis https://github.com/hyPiRion/c-rrb=0A=
=0A=
Aside: An Immutable Map/Set would probably have a stronger argument than im=
mutable sequeunces/deques/linked lists (arrays support copy on write alread=
y so the original array is immutable, but arbitrary keys are something arra=
ys don't support),=0A=
but voters that didn't see a use case for immutables in core might be uncon=
vinced.=0A=
=0A=
Thanks,=0A=
Tyson=0A=