Re: [RFC] Allow specifying keys in list()

On 19 January 2016 at 02:20, Andrea Faulds <ajf@ajf.me> wrote: > After considering how to implement this at the opcode level, I've decided > again that it's not worth it. Mixing keyed and unkeyed elements is not only > an implementation nuisance (it's not necessarily hard, but it makes things > more complicated), it's not something likely to be used in practice, and I > think it's probably an indicator of bad code. > > Now, I can understand your case, but I don't really want to add a special > exception for it. If we're allowing keyed or unkeyed, not both, then we > should stick to it. > > Also, I realised that in your case (list(7 => $a, $b, $c, > $d)), there are > other, possibly cleaner ways to do it, such as: > > list($a, $b, $c, $d) = array_slice($array, 7); > > This way is simple and clear. It's probably clearer than `list(7 => $a, $b, > $c, $d)` in fact. There's also this: > > list(($i = 7) => $a, ++$i => $b, ++$i => $c, ++$i => $d) = $array; > > This way is horrible, but at least demonstrates there are other ways to > achieve what you want. Thanks, I wouldn't worry about it. The use-case was hypothetical anyway. I could see how I might use it, but I have nowhere at present I'd actually use it. list($offset => $a, $b, $b, $c) = $array; $offset += 4; vs. list($a, $b, $c, $d) = array_slice($array, $offset); $offset += 4; Probably not worth the complexity overhead as you suggest, for what amounts to (arguably) syntactic sugar. I also doubt it would see enough usage to warrant the small gain of avoiding the fcall there.