Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:122478
Feedback-ID: i8414410d:Fastmail
User-Agent: Cyrus-JMAP/3.11.0-alpha0-153-g7e3bb84806-fm-20240215.007-g7e3bb848
Precedence: bulk
MIME-Version: 1.0
Message-ID: <36e90d8d-d275-4ce9-9dd9-1e2422c6d3a9@app.fastmail.com>
In-Reply-To: <52C6F501-8E23-42D7-8541-88A22AD79375@koalephant.com>
References: <fca97d6c-d4ec-4000-958a-72c32beb8421@app.fastmail.com>
 <CAJp_myUayeQ=irpFUG0OrUj=PjE3_mx395h_ARSdLmdyEB6_EQ@mail.gmail.com>
 <790b5b4e-f51b-4050-a12a-5fa903d0568f@app.fastmail.com>
 <52C6F501-8E23-42D7-8541-88A22AD79375@koalephant.com>
Date: Fri, 23 Feb 2024 15:58:02 +0000
To: "php internals" <internals@lists.php.net>
Subject: Re: [PHP-DEV] [RFC[ Property accessor hooks, take 2
Content-Type: text/plain
From: larry@garfieldtech.com ("Larry Garfield")

On Fri, Feb 23, 2024, at 8:33 AM, Stephen Reay wrote:
>> On 23 Feb 2024, at 06:56, Larry Garfield <larry@garfieldtech.com> wrote:

> Hi Larry,
>
> It's good to see this idea still progressing.
>
>
> I have to agree with the other comment(s) that the implicit 
> `$field`/`$value` variables seem odd to me. I understand the desire for 
> brevity and the potential future scope of reused hooks, but this 
> concept seems to fly in the face of many years of PHP reducing "magic" 
> like this.

The "magic" that PHP has been removing is mostly weird and illogical type casting.  As noted, neither of these variables are any more "magic" than $this.

However, since it seems no one likes $field, we have removed it from the RFC.  Of note, to respond to your comment further down, $this->{__PROPERTY__} will not work.  The virtual-property detection looks for the AST representation of $this->propName, and only that.  Dynamic versions that turn into that at runtime cannot work, as it needs to be known at compile time.

For $value, however, we feel strongly that having the default there is a necessary part of the ergonomic picture.  In particular, given your comments here:

> To give one answer to your question about ambiguity if the `$value` 
> parameter is required - I don't believe this is actually ambiguous, in 
> the context of PHP:
> - method parameters in child classes don't implicitly 'inherit' the 
> parent method parameter's type if they don't define one (they widen to 
> mixed);
> - method return types have no implicit inheritance, they must declare a 
> compatible return type;
> - typed class properties don't implicitly inherit the parent type when 
> the type left off a child property - they must declare the same type.
>
> AFAIK there is no existing behaviour in PHP where omitting a type would 
> mean "the type is implicitly inherited from X", it either means the 
> same as mixed, or it's an error.

That to me suggests that IF a custom variable name is provided, we should require also specifying the type.  In which case, in the 95% case, if we require the full argument signature then the 95% case would need to double-specify the type, which is a hard-no from an ergonomic standpoint.

Especially combined with the suggestion yesterday to allow return-to-set in the short-set version, that would mean comparing this:

public string $phone {
    set(string $phone) => $this->phone = $this->sanitizePhone($phone);
}

To this:

public string $phone {
    set => $this->sanitizePhone($value);
}

And to me, there's absolutely no contest.  The latter has about 1/3 as many places for me to make a typo repeating the same information over again.  Now imagine comparing the above in a property that's used with constructor promotion.  

public function __construct(
    public string $phone { set(string $phone) => $this->phone = $this->sanitizePhone($phone); }
    public string $phone { set => $this->sanitizePhone($value); }
) {}

Again, it's absolutely no contest for me.  I would detest writing the longer version every time.

If PHP has been moving away from weird and inexplicable magic, it's also been moving away from needless boilerplate.  (Constructor promotion being the best example, but not the only; types themselves are a factor here, as are arrow functions.)  As the whole point of this RFC is to make writing common code easier, requiring redundant boilerplate for it to work is actively counter-productive.

So what I'd suggest instead is "specify the full signature if you want a custom name OR wider type; or omit the param list entirely to get a same-type $value variable, which 99% of the time is all you need."  We get that in return for documenting "$value is the default", which for someone who has already figured out $this, should be a very low effort to learn.

> Also, a small nitpick: The link to your attributeutils repo in the 
> examples page, is broken, and it would be nice to see a few examples 
> showing the explicit version of the hooks.

Link fixed, thanks.  What do you mean explicit version of the hooks?

--Larry Garfield