Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:118881
MIME-Version: 1.0
References: <CADU7aos_g_ZwusqogfsGn33iQ+ZQ3MwW-WdMtSYKn5RE+2=NOQ@mail.gmail.com>
In-Reply-To: <CADU7aos_g_ZwusqogfsGn33iQ+ZQ3MwW-WdMtSYKn5RE+2=NOQ@mail.gmail.com>
Reply-To: Levi Morrison <levi.morrison@datadoghq.com>
Date: Mon, 24 Oct 2022 21:39:30 -0600
Message-ID: <CAAKU0ukJ-6iCFTyL8R7ug4QrF0WjYEdJaAzq6Pxy1ZkGKWHtrw@mail.gmail.com>
To: =?UTF-8?Q?K=C3=A9vin_Dunglas?= <kevin@dunglas.fr>
Cc: internals@lists.php.net
Content-Type: text/plain; charset="UTF-8"
Subject: Re: [PHP-DEV] Proposal to incrementally improve timeout and signal handling
From: internals@lists.php.net ("Levi Morrison via internals")

This kind of thing requires careful thought. My first glance through
it looks good and I definitely want PHP to stop abusing the SIGPROF
signal because it prevents usage by actual profilers, which is
annoying. Full disclosure: I write a profiler for my job.

In addition, I think that there should probably be an internals API
for handing out POSIX realtime signal numbers for platforms which
support it. Quoting from the signal man page:

> Unlike standard signals, real-time signals have no predefined
> meanings: the entire set of real-time signals can be used for
> application-defined purposes.

One could interpret this as meaning we should also make it accessible
to userland, but in line with the one step at a time approach, we can
do that later if we decide we need to. But this is something an
extension can't choose to do by itself: it must be done in core.

I'll review the proposal sometime more carefully. Thanks for bring it up.