Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:70627
Mailing-List: contact internals-help@lists.php.net; run by ezmlm
Received-SPF: pass (pb1.pair.com: domain ajf.me designates 198.187.29.233 as permitted sender)
Message-ID: <52AB3AAE.60403@ajf.me>
Date: Fri, 13 Dec 2013 16:49:50 +0000
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:24.0) Gecko/20100101 Thunderbird/24.2.0
MIME-Version: 1.0
To: Ferenc Kovacs <tyra3l@gmail.com>
CC: PHP internals <internals@lists.php.net>
References: <52AA1649.2090601@ajf.me> <CAH-PCH5sr=zS92fM_7e6HWsTxgQjJw04YcAcTO+F+YTtFvNc4w@mail.gmail.com>
In-Reply-To: <CAH-PCH5sr=zS92fM_7e6HWsTxgQjJw04YcAcTO+F+YTtFvNc4w@mail.gmail.com>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit
Subject: Re: [PHP-DEV] [PROPOSAL] Alphanumeric decrement
From: ajf@ajf.me (Andrea Faulds)



On 13/12/13 10:02, Ferenc Kovacs wrote:
>
> So changing the behavior in my opinion would be a BC break, and given
> the nature of this feature, I think that there are no solution, which
> would eliminate all the edge-cases and make the feature truly
> consistant/symetric to the current incrementing and to our type
> juggling, so I think we can't even call it a bugfix(and ship it in a
> minor), when we change it from mostly-broken to slightly-broken.
>

It is technically a BC-break if something relies on being able to 
infinitely decrement alphanumerics and have the same result, but I doubt 
any code actually relies on it.

I think this entirely consistent with incrementing at present, with the 
one exception of "a" decrementing to zero. Making "a" not decrement at 
all is a compromise I might consider.

At risk of derailing the thread, I think the current "no BC breaks in 
minors" policy is somewhat misguided. The problem is that a lot of 
bugfix are technically a BC break (unless code is supposed to be 
consistent with how PHP's supposed to work and not how it actually 
works), and in practise you'll have to introduce some BC breaks when 
adding new features (for example, making certain constant names keywords 
would break existing code which used those names). Hence perhaps we 
should change the rules to say that certain BC breaks are allowed, but 
only in very limited and specific circumstances.

-- 
Andrea Faulds
http://ajf.me/