375882 - Decompiler still outputs "NaN" (etc) incorrectly for |case|

Status: VERIFIED FIXED

(Core :: JavaScript Engine, defect, P2)

Description

[Jesse Ruderman]

Bug 351219 didn't fix switch |case| decompilation:

js> function() { switch(a) { case 0/0: a; case 1/0: b; case -1/0: c; case -0: d; } }
function () {
    switch (a) {
      case NaN:
        a;
      case Infinity:
        b;
      case -Infinity:
        c;
      case 0:
        d;
      default:;
    }
}

Comment 1

[Brendan Eich [:brendan]]

Attachment: fix

Comment 2

[Brendan Eich [:brendan]]

Attachment: fix, v2

Forgot static, and added an assertion.

/be

Comment 3

[Brendan Eich [:brendan]]

Attachment: patch to commit

Use ?: in JSOP_NUMBER's decompiler instead of if/else with common left-hand side (todo) of single assignment statement in then and else clauses. Bonus: no Sprint* function calls have to wrap unnaturally given the wrapped ?:.

/be

Comment 4

[Brendan Eich [:brendan]]

Fixed on trunk:

js/src/jsopcode.c 3.222

/be

Comment 5

[Bob Clary [:bc] (inactive)]

/cvsroot/mozilla/js/tests/js1_5/decompilation/regress-375882.js,v  <--  regress-375882.js
initial revision: 1.1

the case 1 / -0 : just looks weird to me.

function ( ) { 
    switch ( a ) { 
        case 0 / 0 : a ; 
        case 1 / 0 : b ; 
        case 1 / - 0 : c ; 
        case - 0 : d ; 
        default : ; 
    } 
}

Comment 6

[Brendan Eich [:brendan]]

We could use -1 / 0 since the expression will be auto-parenthesized as needed, based on operator precedence. File a new bug for this prettier form if you like?

/be

Comment 7

[Bob Clary [:bc] (inactive)]

verified fixed linux, windows, mac* shell 20070406