Re: Revisiting RFC: Engin e Warnings -- Undefined array index

On 12 December 2022 22:20:27 GMT, Dan Liebner <dliebner@gmail.com> wrote: >Here's my biggest criticism: the change essentially forces this: >$arr['key'] > >to be rewritten as this (or some other intolerably bad alternative): >isset($arr) && isset($arr['key']) && $arr['key'] Fortunately, this is not the case, the actual replacement code is this: $arr['key'] ?? null There has previously been a suggestion of introducing an even shorter syntax, such as this: $arr[?'key'] (With the difference that only one key is marked optional, whereas ?? allows any number of dimensions - $foo['bar']['baz'] ?? null would be equivalent to $foo[?'bar'][?'baz'].) This would avoid the inconsistency of configurable behaviour, and allow users to make clear with a single keystroke that they know a key might be missing, while giving them useful feedback for typos and other mistakes that lead to a key being unexpectedly missing. I think this is a general approach that should be more explicitly embraced: introduce convenient syntax for valid use cases, while making accidents obvious. Regards, Hi Dan, -- Rowan Tommins [IMSoP]