Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:128977
Precedence: list
MIME-Version: 1.0
References: <CAMW7n8DGbFnQzjt3s9CcOBwhp+MUOV+K3U=yY=ULOTdUXaOLHQ@mail.gmail.com>
In-Reply-To: <CAMW7n8DGbFnQzjt3s9CcOBwhp+MUOV+K3U=yY=ULOTdUXaOLHQ@mail.gmail.com>
Date: Mon, 27 Oct 2025 15:01:41 +0100
Message-ID: <CAEKnhAGxuP4VKAc-LLW-OKWaLWS8KZDr46=eaDPk4FMsxa_KSg@mail.gmail.com>
Subject: Re: [PHP-DEV] PHP True Async RFC Stage 4
To: Edmond Dantes <edmond.ht@gmail.com>
Cc: php internals <internals@lists.php.net>
Content-Type: multipart/alternative; boundary="0000000000006eb6af06422457f0"
From: bukka@php.net (Jakub Zelenka)

--0000000000006eb6af06422457f0
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

Hi,

On Sun, Oct 5, 2025 at 7:26=E2=80=AFAM Edmond Dantes <edmond.ht@gmail.com> =
wrote:

> Good day, everyone. I hope you're doing well.
>
> I=E2=80=99m happy to present the fourth version of the RFC. It wasn=E2=80=
=99t just me
> who worked on it =E2=80=94 members of the PHP community contributed as we=
ll.
> Many thanks to everyone for your input!
>
> https://wiki.php.net/rfc/true_async


I just re-read it again with all the feedback provided and I think it
should get further stripped as it seems problematic to get an agreement on
all of this and properly discuss it on ML all the details.

I think this (or more v5) should strip the following:

- exposing Scope and all operations in it. It means it should allow using
only the default scope in this version. That should significantly reduce
the size of the RFC as it removes structured concurrency and other parts
related to scopes (including the reduction of error handling logic).
- timer functions could also be removed even though it will make it less
usable but the point is to make it as small as possible and those are not
absolutely essential parts.
- critical section should be stripped as well
- nginx unit example should be removed as it might be confusing - I
understand why it was added but it might be more confusing than useful
- drop php.ini setting and just use default for now

The idea is to make this as small as possible so this might be possible to
discuss and get actually more people to read the RFC (this is just too
long). This will also allow to concentrate on specific pieces like for
example deciding whether to use FutureLike or Awaitable.

We just had a chat about this work during our PHP Foundation and the
agreement seems to be that this should be reduced and come in smaller
pieces to be able to better figure out what's actually useful for PHP.
Another point was also to make clear that this proposal is not meant to
introduce a new "right" way to use PHP but it's actually useful everywhere.
We actually discussed this with Edmond privately and agreed that this is
useful for FPM as well and he even created an example proving it. So we
should just try to make it clearer in the RFC as there was some confusion
in the discussion.

Kind regards,

Jakub

--0000000000006eb6af06422457f0
Content-Type: text/html; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div dir=3D"ltr"><div>Hi,</div><br><div class=3D"gmail_quo=
te"><div dir=3D"ltr" class=3D"gmail_attr">On Sun, Oct 5, 2025 at 7:26=E2=80=
=AFAM Edmond Dantes &lt;<a href=3D"mailto:edmond.ht@gmail.com" target=3D"_b=
lank">edmond.ht@gmail.com</a>&gt; wrote:<br></div><blockquote class=3D"gmai=
l_quote" style=3D"margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,20=
4,204);padding-left:1ex">Good day, everyone. I hope you&#39;re doing well.<=
br>
<br>
I=E2=80=99m happy to present the fourth version of the RFC. It wasn=E2=80=
=99t just me<br>
who worked on it =E2=80=94 members of the PHP community contributed as well=
.<br>
Many thanks to everyone for your input!<br>
<br>
<a href=3D"https://wiki.php.net/rfc/true_async" rel=3D"noreferrer" target=
=3D"_blank">https://wiki.php.net/rfc/true_async</a></blockquote><div><br></=
div><div>I just re-read it again with all the feedback provided and I think=
 it should get further stripped as it seems problematic to get an agreement=
 on all of this and properly discuss it on ML all the details.</div><div><b=
r></div><div>I think this (or more v5) should strip the following:<br><br><=
/div><div>- exposing Scope and all operations in it. It means it should all=
ow using only the default scope in this version. That should significantly =
reduce the size of the RFC as it removes structured concurrency and other p=
arts related to scopes (including the reduction of error handling logic).</=
div><div>- timer functions could also be removed even though it will make i=
t less usable but the point is to make it as small as possible and those ar=
e not absolutely essential parts.</div><div>- critical section should be st=
ripped as well</div><div>- nginx unit example should be removed as it might=
 be confusing - I understand why it was added but it might be more confusin=
g than useful</div><div>- drop php.ini setting and just use default for now=
</div><div>=C2=A0</div><div>The idea is to make this as small as possible s=
o this might be possible to discuss and get actually more people to read th=
e RFC (this is just too long). This will also allow to concentrate on speci=
fic pieces like for example deciding whether to use FutureLike or Awaitable=
.</div><div><br></div><div>We just had a chat about this work during our PH=
P Foundation and the agreement seems to be that this should be reduced and =
come in smaller pieces to be able to better figure out what&#39;s actually =
useful for PHP. Another point was also to make clear that this proposal is =
not meant to introduce a new &quot;right&quot; way to use PHP but it&#39;s =
actually useful everywhere. We actually discussed this with Edmond privatel=
y and agreed that this is useful for FPM as well and he even created an exa=
mple proving it. So we should just try to make it clearer in the RFC as the=
re was some confusion in the discussion.</div><div><br></div><div>Kind rega=
rds,</div><div><br></div><div>Jakub</div><div><br></div></div></div>
</div>

--0000000000006eb6af06422457f0--