Switch statement without reached the case block

[igor-simoes]

Hi everyone,
there is an inconsistency in Chakra if a switch statement with only one case block evaluation that could be not reached.

version: 1.10.2
OS: Ubuntu 16.04 x64

steps to reproduce:

{ function f() { return "f 0"; } }
switch (1)
{
  case /0.0/gi:
    function f() {
       return "f 1";
    }
    break;
}
print( f() );

Actual results:
f 1

Expected results:
f 0

if we try to add the default block (default: or default: break;), the same occurs as above.

V8 and SpiderMonkey works as expected.

cinfuzz

[rhuanjl]

The issue here is that the function definition is being hoisted to the top of the switch statement out of the case that contains it - this doesn't occur if you wrap the contents of the case in { }.

I don't actually know the spec on this to know what behaviour is right - should a case clause that isn't wrapped in { } still be treated as a different block?

[rhuanjl]

@zenparsing or @chicoxyzzy do either of you know the correct answer per spec here? I tried reading through the spec but can't work it out.

If you have a case block WITHOUT { } should function definitions be hoisted out of it into the block of the switch statement that contains them?

[chicoxyzzy]

I think that this is the same reason as in #5511 (Annex B 3.3 is not implemented or implemented incorrectly)