Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:83123
Mailing-List: contact internals-help@lists.php.net; run by ezmlm
Received-SPF: pass (pb1.pair.com: domain gmail.com designates 209.85.213.52 as permitted sender)
MIME-Version: 1.0
In-Reply-To: <CAAyV7nGjdguX=4Nt0H=kYnwg8rZ1BkWf8duGZfzheOXXV61pqA@mail.gmail.com>
References: <CAAyV7nGjdguX=4Nt0H=kYnwg8rZ1BkWf8duGZfzheOXXV61pqA@mail.gmail.com>
Date: Wed, 18 Feb 2015 21:23:58 +0000
Message-ID: <CAPwsocF55ZgYXr+QdP-vny4qVApoNm7aE6=NCJME=uSUBzZLqw@mail.gmail.com>
To: Anthony Ferrara <ircmaxell@gmail.com>
Cc: "internals@lists.php.net" <internals@lists.php.net>
Content-Type: text/plain; charset=UTF-8
Subject: Re: [PHP-DEV] [RFC-Discuss] Scalar Type Declarations v0.5
From: leight@gmail.com (Leigh)

On 18 February 2015 at 20:44, Anthony Ferrara <ircmaxell@gmail.com> wrote:
> Dear Internals,
>
> Since the resignation of Andrea, the mixed-mode type hint (called
> declaration in the proposal) proposal has been left abandoned.
> Considering that the ending votes were 67/34 (66.3%) with several
> no-votes being only due to reasonably minor issues with the proposal,
> I would like to re-propose her RFC with three minor modifications:
>
> 1. declare(strict_types=1) (if used) is required to be the first
> instruction in the file only. No other usages allowed.
> 2. declare(strict_types=1) {} (block mode) is specifically disallowed.
> 3. int typed variables can resolve a parameter type of float So
> calling requiresAFloat(10) will work even in strict mode.
>
> As this topic has and is being discussed to death, I have put a very
> large "discussion points" section:
> https://wiki.php.net/rfc/scalar_type_hints_v5#discussion_points
>
> I would kindly ask, before replying that you check to see if your
> question is answered in that list.

General comments from me, as a radical strictist. No need to reply and
tell me I'm "wrong", these are my opinions.

> Internal Functions Like ceil() Return Unexpected Types

My opinion is that functions should return sane types for their
intended purpose, and functions that do not should be "fixed".

> Integers Should Be Accepted For Strict float Arguments
> Int->Float Exception Makes Strict Mode "Flawed"

My opinion is that strict should mean strict.

> Static Analysis Is Possible With Weak Declarations

The advocacy of allowing `accepts_float(returns_int());` doesn't help
the cause of static analysers in strict mode.

> Why Not Add Support For Null?

I agree a null type is pointless, however I would like to raise the
discussion point of existing hinted optional parameters being allowed
a default of null. i.e. fn(Obj $o = null) {}

Do you think this is acceptable for any other types? I can only think
of it being useful if resource and a generic object type are added.
but others may disagree. (Ideally the resource type will go away over
time and will be replaced with objects like GMP anyway :))

Still no mention of a way to enable strict by default. So lets try
some different rationale. If I personally want to develop in strict
mode, I can do that with this proposal, I can add a declare to the top
of every file and be as strict as I like. However not everyone else is
going to want to use my code in strict mode, so they will have to go
and remove some/all of those strict declarations. (Lets take callbacks
being evaluated in the context they are called rather than the one
they are created in). I would like a way of enabling strict by
default, immutable to scripts so that users cannot be forced into this
mode, and lets the radicals and the weaklings* play together in
harmony.

For the rest of the RFC, I either agree with or have no strong
opinions about the points raised.


*it's a joke you guys.