Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:111458
MIME-Version: 1.0
References: <6cfb77d4d511b5c591e7d7b5cc80199955123b2cd4276a5bca2327e4b02942db@mahalux.com>
 <CA+kxMuTSDVJFXL=Cfx3+XQLpF60ajcSdVy-N5eXsfL43iujjTw@mail.gmail.com>
 <c99b01d42ba5e34e4503b16a84f30d4955b0acc398fb567aeec84f585dbc4bf9@mahalux.com>
 <CAFPFaM+B__GYBxLMsfVKiPefOhCHwMWCVv1ccXoMPCeD7W+dnA@mail.gmail.com>
In-Reply-To: <CAFPFaM+B__GYBxLMsfVKiPefOhCHwMWCVv1ccXoMPCeD7W+dnA@mail.gmail.com>
Date: Tue, 11 Aug 2020 13:13:13 +0100
Message-ID: <CA+kxMuS7=E8x1YnDXV5wOWHzirzYUZro8BVGoLrc1Gp_aGoigA@mail.gmail.com>
To: "G. P. B." <george.banyard@gmail.com>
Cc: =?UTF-8?B?TWljaGFlbCBWb8WZw63FoWVrIC0gxIxWVVQgRkVM?= <vorismi3@fel.cvut.cz>, 
	PHP internals <internals@lists.php.net>, =?UTF-8?Q?Bj=C3=B6rn_Larsson?= <bjorn.x.larsson@telia.com>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Subject: Re: [PHP-DEV] Allow sleep() to accept non-integer values
From: Danack@basereality.com (Dan Ackroyd)

Bj=C3=B6rn Larsson wrote:
> Given this unexpected behaviour, one could almost see it as a bug.

This isn't a suddenly noticed new bug. That code has worked like that
since the sleep function was committed twenty-two years ago or for
five years since the release of PHP 7 and the weak/strict RFC
continued the "if you don't want float -> int conversions, use strict
mode" choice.

There are quite a few functions in PHP that take int parameters, that
have this behaviour.

These could all be 'fixed' by people using strict mode, and you could
make a good argument for making strict mode the default, but it's
probably a bit late in the 8.0 release cycle for that, and it _might_
be  a little contentious.

I agree that the PHP standard library is not as good as it could be,
but tweaking it piece-by-piece is not a good way of fixing it.

Doing it like that would lead to subtle BC breaks that are not obvious.

G.P.B. wrote:
> I'd rather see the failure condition changed to reject 0 or add a
> E_NOTICE/E_WARNING if 0 is passed as this has questionable semantics.

sleep(0) is a useful thing to do in some circumstances. It has the
same effect as Thread.yield() in Java.

Or in English, it allows a process to say "Hello OS scheduler, I don't
want to sleep right now, but if there are any other threads that are
waiting for their turn, now is a great time to suspend my execution,
as I'm not in the middle of anything, otherwise I'll keep processing."

Again, yes we need a better way of evolving the core libraries. But
tweaking it on a function by function basis is not a good way of doing
that.

cheers
Dan
Ack