Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:84233
Mailing-List: contact internals-help@lists.php.net; run by ezmlm
Received-SPF: error (pb1.pair.com: domain kelunik.com from 81.169.146.162 cause and error)
MIME-Version: 1.0
In-Reply-To: <54F5A8C4.8070508@gmail.com>
References: <CAH8MpnmTj8Y5cST2M0LQZFKiV6sp_sPV=h8dku3_HPxsQYeYOg@mail.gmail.com>
	<CAJ26g5SXGhFB6OQqbe7NoWC7sNgzeyhvjrAVXfrH21hgDhS_=A@mail.gmail.com>
	<CAH8Mpnmw8eC6YOk6M6D-PoP-2zY20_GzAEgZRLBiA9Uw1yL-4Q@mail.gmail.com>
	<CAH8Mpnnin64H94wYDX_xQTamkZLQjxkCbkytf4C5+sfq+DguLg@mail.gmail.com>
	<CANUQDCih62xnJHPZ9rHNOsKppmJDR=zvuEYya0knsH4+9TE7=Q@mail.gmail.com>
	<CAH8MpnmWww+yXf9wi7t5m-KCstdbyKmdayU7hDHmrkD09=67Jw@mail.gmail.com>
	<CALKiJKqCy4VdPXnuHUzoaBF+9iy1+Qg19mnujSPNtm7=KD6u4g@mail.gmail.com>
	<CANUQDCitouszxLC-qYLga1WRA0gQFj97D1w62jJYn+82wOUzDA@mail.gmail.com>
	<54F59C45.7080704@gmail.com>
	<BLU436-SMTP1398948B1BF592EE485D81AE2110@phx.gbl>
	<54F5A8C4.8070508@gmail.com>
Date: Tue, 3 Mar 2015 13:52:21 +0100
Message-ID: <CANUQDCj46uNwPnwa8ezV5A1dJW+ygH9hnJooYFhxMDZGbWUMUg@mail.gmail.com>
To: Rowan Collins <rowan.collins@gmail.com>
Cc: Bob Weinand <bobwei9@hotmail.com>, Daniel Lowrey <rdlowrey@php.net>, 
	"internals@lists.php.net" <internals@lists.php.net>
Content-Type: multipart/alternative; boundary=089e01493f60381b2a051061cef4
Subject: Re: [PHP-DEV] [RFC Discuss] Generator Delegation
From: me@kelunik.com (Niklas Keller)

--089e01493f60381b2a051061cef4
Content-Type: text/plain; charset=UTF-8

2015-03-03 13:27 GMT+01:00 Rowan Collins <rowan.collins@gmail.com>:

> Bob Weinand wrote on 03/03/2015 12:08:
>
>  Am 03.03.2015 um 12:34 schrieb Rowan Collins <rowan.collins@gmail.com>:
>>>
>>> Niklas Keller wrote on 03/03/2015 10:55:
>>>
>>>>     Gr, top-posting...
>>>>
>>>>
>>>> Sorry, was on mobile. ;-)
>>>>
>>>>     However, since the existence of the word "yield" is the only thing
>>>>     that
>>>>     marks a coroutine now, how about using a variant of that for the
>>>> final
>>>>     value, e.g. "yield final $foo"?
>>>>
>>>>
>>>> What's the final value? The last "yield"ed value or a return?
>>>>
>>> "yield final" would mark the final result of the coroutine, as opposed
>>> to the intermediate values passed out with a normal "yield".
>>>
>>>
>>>  Just to give you some real world example:
>>>> If you're using "return", it would look like that:
>>>>
>>>> public function getSession ($sessionId) {
>>>>     $result = yield $this->redis->get("session.{$sessionId}")); //
>>>> We're waiting here until redis responded.
>>>>     return json_decode($result);
>>>> }
>>>>
>>> My suggestion is simply to change the keyword:
>>>
>>> public function getSession ($sessionId) {
>>>     $result = yield $this->redis->get("session.{$sessionId}")); //
>>> We're waiting here until redis responded.
>>>     yield final json_decode($result);
>>> }
>>>
>>>
>>> The reasoning being that when you run getSession(42), it *doesn't*
>>> return the result of json_decode(), it returns a pointer to the coroutine's
>>> state, which you can resume later.
>>>
>>> Actually, I don't think that example makes sense, because JSON gets sent
>>> out at the first yield, and then sent back in, so the caller would look
>>> something like this:
>>>
>>> $foo = getSession(42);
>>> $json_data = $foo->current();
>>> $foo->send($json_data);
>>> $decoded = $foo->getReturn();
>>>
>>> But never mind, I think we both get the idea.
>>>
>>> I understand the desire for a "final result", but I don't like reusing
>>> the word "return", because it's never "returned" as the result of running
>>> the function, it's just made available through some specific method/syntax.
>>>
>>> Regards,
>>> --
>>> Rowan Collins
>>> [IMSoP]
>>>
>> Hey,
>>
>> We currently already have "return;" (without value) as allowed syntax to
>> terminate Generator execution. Thus, the logical consequence is just adding
>> a value to that return.
>>
>
> Hm, I didn't realise that. I suppose it makes sense that you'd want a way
> to do that, but it does make it very hard to see if something's going to
> behave as a function or a coroutine if they use the same keywords to mean
> different things.
>
>  When we want to *return*, it should be also a real *return". Yes, it is
>> just accessible through special syntax (or a method), but it's still
>> *return*ing into the calling frame.
>>
>
> Well, so is "yield" - it just leaves the Generator in a state that can be
> resumed, whereas "return"/"yield final" leaves it in a final, non-resumable
> state.
>
>  Why should the word "return" be unique to methods or functions?
>>
>
> It just doesn't feel like the same thing as a return value to me, for the
> same reason a generator doesn't feel like the same thing as a function. In
> "function foo() { return 42; }", "return" means "this is what you'll get
> when you run foo()"; in "function foo() { return 42; yield; }", what you
> get when you run foo() is a pointer to the resumable state, and return
> means "this is what you'll get if you ask the generator/coroutine instance
> you get by running foo() for its final result".
>
>
> Regards,
> --
> Rowan Collins
> [IMSoP]
>

It's still different from "return". You said, it "mark[s] the final
result", so I'd expect that finally yielded value would be at the iterator,
but it's not. It's separate.

How would you call the method then, that makes that "yield final" value
available?

Regards, Niklas

--089e01493f60381b2a051061cef4--