Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:112466
To: internals@lists.php.net
References: <d1c4689e-9120-4e90-837c-a1b3eac63cd6@www.fastmail.com>
 <CAGgaK7KucDd8VZ-5OPsN1jg_7LTJLY8Btoa7XbXdrXHMk1yE_A@mail.gmail.com>
Message-ID: <a4be88c4-eac6-d0b8-506a-76ad3c8d7c56@gmail.com>
Date: Tue, 8 Dec 2020 09:11:05 +0000
User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:78.0) Gecko/20100101
 Thunderbird/78.5.1
MIME-Version: 1.0
In-Reply-To: <CAGgaK7KucDd8VZ-5OPsN1jg_7LTJLY8Btoa7XbXdrXHMk1yE_A@mail.gmail.com>
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Transfer-Encoding: 8bit
Content-Language: en-GB
Subject: Re: [PHP-DEV] [RFC] Enumerations
From: rowan.collins@gmail.com (Rowan Tommins)

On 07/12/2020 20:20, Max Semenik wrote:

> I myself have tried to draft my proposal athttps://wiki.php.net/rfc/enum_v2  which has
> aspects both similar and different from yours.


I'm afraid that proposal would be a strong -1 from me, as "fancy 
constants" are my least favourite type of enum. In particular, this 
would be a deal-breaker:


> Conversion from other types is not checked, thus enums can hold values 
> not covered by their constants. 


The main reason I want enums is to make it impossible to represent 
invalid states, so that they don't have to be accounted for at run-time. 
Allowing someone to write (Month)13 without an immediate error would 
completely invalidate that purpose.



> * Performance concern: I suspect that most developers will not use the
> advanced features from this proposal and your future plans much, instead
> they will want to use it in very simple contexts, e.g.
> "$order->calculateWeight(Weight::Gross)". Would every such call require an
> object allocation and initialization? What about comparisons, would
> expressions like "if ($cardType === Suit::Clubs) ..." require a new object?


As I understand it, Weight::Gross just looks up the same object each 
time, so no allocation would be needed unless this happened to be the 
first time you'd mentioned that enum.

Once you have an instance of an object, you're just passing around or 
comparing a pointer (i.e. an integer) anyway, so I would expect it to 
have roughly the same performance as passing or comparing a bare integer.


Regards,

-- 
Rowan Tommins (né Collins)
[IMSoP]