Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:129395
Feedback-ID: ifab94697:Fastmail
Precedence: list
MIME-Version: 1.0
Date: Sat, 22 Nov 2025 15:37:19 +0100
To: "Edmond Dantes" <edmond.ht@gmail.com>, "Thomas Bley" <mails@thomasbley.de>
Cc: internals@lists.php.net
Message-ID: <2e7ab8cc-a304-445c-9ef2-0449e6ce7873@app.fastmail.com>
In-Reply-To: 
 <CAMW7n8AycfX4NMfjJ+GzFsPqRaSae-t7Cm3H5z8apRzXUgaVZw@mail.gmail.com>
References: 
 <CAMW7n8D5B=XVX_Jes+LV=2fPwjQr_RBVphPi_Cxm8-_Rwpdo_Q@mail.gmail.com>
 <CADK1yX+jGtqQVYFTo5YmkenPrNcXz7Dd7RCUKWGrq+50pnKOnw@mail.gmail.com>
 <CAMW7n8Dogjj2JFJBNArk-BfvOHRqHKMAFReXJAhDm2UPV1TF7A@mail.gmail.com>
 <CADK1yXKgkuKmOMB7KA9GNQMtpNw7ZgWcPEPE_Uhr_+8eReDCEA@mail.gmail.com>
 <CAMW7n8D+RGAFDY3oGnzaAabrhW9TOMeBgtvx4zCy3AoBUb5szA@mail.gmail.com>
 <CAEKnhAFsXpfZy8FQyA4Z9bNyNa26rCCn5GO=oQ274e1K2+GkAw@mail.gmail.com>
 <E4EE97C5-76EB-4C65-A3BE-7FFE27DD1500@rwec.co.uk>
 <383528143.397454.1763760932585@email.ionos.de>
 <CAMW7n8BjVM7SdM8Jn3dOmrXgngxycerZr9Cm4jeACnXaETQCWQ@mail.gmail.com>
 <691963327.399484.1763796022993@email.ionos.de>
 <CAMW7n8DnE_M3MOOfH_4x9DOZyQyT6FCbhBtU+6tuDD3Rw0vzbQ@mail.gmail.com>
 <1402906904.784903.1763811759896@email.ionos.de>
 <2015208655.784984.1763811978476@email.ionos.de>
 <CAMW7n8D0=6hEPpZpGe2SrYKqgFqsDkPAjHBP4g5K45bFPaiNdw@mail.gmail.com>
 <980296152.1071.1763813573083@email.ionos.de>
 <CAMW7n8DXA501uGOY3952y8YhAJDVRrPYEiuw+Wzc5=oEciLVZQ@mail.gmail.com>
 <92865666.4510.1763818506332@email.ionos.de>
 <CAMW7n8AycfX4NMfjJ+GzFsPqRaSae-t7Cm3H5z8apRzXUgaVZw@mail.gmail.com>
Subject: Re: [PHP-DEV] [VOTE] True Async RFC 1.6
Content-Type: multipart/alternative;
 boundary=aa7693f56e6048e093b20cc7330f6cde
From: rob@bottled.codes ("Rob Landers")

--aa7693f56e6048e093b20cc7330f6cde
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable



On Sat, Nov 22, 2025, at 14:55, Edmond Dantes wrote:
> > So I guess you want to use spawn() in a similar way as call_user_fun=
c() works.
> yes
>=20
> > This changes the behavior of file_get_contents() from the outside
> No.
>=20
> ```php
> function file_get_contents(string $filename): string
> {
>     $fh =3D fopen();
>=20
>     // It creates an EPOLL event so it can wake us when the data
> becomes available.
>      $event =3D ReactorAPI.create_event_from($fh);
>     $waker =3D Scheduler.getCurrentWaker();
>     // Event Driven logic inside.
>     $waker.add_event($event, function() use($waker) {
>           // Wakeup this coroutine
>           $waker.wake();
>      });
>=20
>     // suspend current coroutine
>     // zz..... z.....
>     Scheduler.suspend();
>=20
>     // Continue here after the IO event
>=20
>     // Now we have date, return
>     return fread($fh, ....);
> }
> ```
>=20
> This is pseudocode. You can assume it always works.
> If you call `file_get_contents` directly, it behaves the same way.
> So it does not matter where `file_get_contents` is called.
> Since all PHP code together with TrueAsync runs inside coroutines,
> `file_get_contents` will suspend the coroutine in which it was invoked.
>=20
> When you call `spawn`, you simply run the function in another
> coroutine, not in your own. But `spawn` has no effect on
> `file_get_contents`.
>=20
> We=E2=80=99re not at risk of DataRace yet :) We don=E2=80=99t have mul=
tithreading.
> And most likely it won=E2=80=99t appear anytime soon.
>=20

We are in data-race territory though:

spawn(fn() =3D> file_put_contents('file', 'long string'))
spawn(fn() =3D> file_get_contents('file'))

I=E2=80=99m on a phone, so I=E2=80=99m not sure I got all the syntax rig=
ht, but hopefully my intent is clear. But if these run =E2=80=9Cconcurre=
ntly=E2=80=9D the scheduler will theoretically batch the reading/writing=
 of bytes in an interleaved way, causing absolute chaos and corruption.

The same thing would happen with db drivers that typically use a token t=
o keep track of which response goes to which request (I maintain an asyn=
c db driver for amphp). There will be a need to ensure the stream cannot=
 be interleaved with other coroutines. I can do this with amphp locks, b=
ut there isn=E2=80=99t even a semaphore implementation to build a lock a=
round.=20

=E2=80=94 Rob
--aa7693f56e6048e093b20cc7330f6cde
Content-Type: text/html; charset=utf-8
Content-Transfer-Encoding: quoted-printable

<!DOCTYPE html><html><head><title></title></head><body><div><br></div><d=
iv><br></div><div>On Sat, Nov 22, 2025, at 14:55, Edmond Dantes wrote:</=
div><blockquote type=3D"cite" id=3D"qt" style=3D""><div>&gt; So I guess =
you want to use spawn() in a similar way as call_user_func() works.</div=
><div>yes</div><div><br></div><div>&gt; This changes the behavior of fil=
e_get_contents() from the outside</div><div>No.</div><div><br></div><div=
>```php</div><div>function file_get_contents(string $filename): string</=
div><div>{</div><div>&nbsp;&nbsp;&nbsp; $fh =3D fopen();</div><div><br><=
/div><div>&nbsp;&nbsp;&nbsp; // It creates an EPOLL event so it can wake=
 us when the data</div><div>becomes available.</div><div>&nbsp;&nbsp;&nb=
sp;&nbsp; $event =3D&nbsp;<a href=3D"http://ReactorAPI.create">ReactorAP=
I.create</a>_event_from($fh);</div><div>&nbsp;&nbsp;&nbsp; $waker =3D&nb=
sp;<a href=3D"http://Scheduler.getCurrentWaker">Scheduler.getCurrentWake=
r</a>();</div><div>&nbsp;&nbsp;&nbsp; // Event Driven logic inside.</div=
><div>&nbsp;&nbsp;&nbsp; $<a href=3D"http://waker.add">waker.add</a>_eve=
nt($event, function() use($waker) {</div><div>&nbsp;&nbsp;&nbsp;&nbsp;&n=
bsp;&nbsp;&nbsp;&nbsp;&nbsp; // Wakeup this coroutine</div><div>&nbsp;&n=
bsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; $<a href=3D"http://waker.=
wake">waker.wake</a>();</div><div>&nbsp;&nbsp;&nbsp;&nbsp; });</div><div=
><br></div><div>&nbsp;&nbsp;&nbsp; // suspend current coroutine</div><di=
v>&nbsp;&nbsp;&nbsp; // zz..... z.....</div><div>&nbsp;&nbsp;&nbsp;&nbsp=
;<a href=3D"http://Scheduler.suspend">Scheduler.suspend</a>();</div><div=
><br></div><div>&nbsp;&nbsp;&nbsp; // Continue here after the IO event</=
div><div><br></div><div>&nbsp;&nbsp;&nbsp; // Now we have date, return</=
div><div>&nbsp;&nbsp;&nbsp; return fread($fh, ....);</div><div>}</div><d=
iv>```</div><div><br></div><div>This is pseudocode. You can assume it al=
ways works.</div><div>If you call `file_get_contents` directly, it behav=
es the same way.</div><div>So it does not matter where `file_get_content=
s` is called.</div><div>Since all PHP code together with TrueAsync runs =
inside coroutines,</div><div>`file_get_contents` will suspend the corout=
ine in which it was invoked.</div><div><br></div><div>When you call `spa=
wn`, you simply run the function in another</div><div>coroutine, not in =
your own. But `spawn` has no effect on</div><div>`file_get_contents`.</d=
iv><div><br></div><div>We=E2=80=99re not at risk of DataRace yet :) We d=
on=E2=80=99t have multithreading.</div><div>And most likely it won=E2=80=
=99t appear anytime soon.</div><div><br></div></blockquote><div><br></di=
v><div>We are in data-race territory though:</div><div><br></div><div>sp=
awn(fn() =3D&gt; file_put_contents('file', 'long string'))</div><div>spa=
wn(fn() =3D&gt; file_get_contents('file'))</div><div><br></div><div>I=E2=
=80=99m on a phone, so I=E2=80=99m not sure I got all the syntax right, =
but hopefully my intent is clear. But if these run =E2=80=9Cconcurrently=
=E2=80=9D the scheduler will theoretically batch the reading/writing of =
bytes in an interleaved way, causing absolute chaos and corruption.</div=
><div><br></div><div>The same thing would happen with db drivers that ty=
pically use a token to keep track of which response goes to which reques=
t (I maintain an async db driver for amphp). There will be a need to ens=
ure the stream cannot be interleaved with other coroutines. I can do thi=
s with amphp locks, but there isn=E2=80=99t even a semaphore implementat=
ion to build a lock around.&nbsp;</div><div><br></div><div id=3D"sig1212=
29152">=E2=80=94 Rob</div></body></html>
--aa7693f56e6048e093b20cc7330f6cde--