This is a continuation of bug 349956. > function () { <x/>[@[*]]; } function () { <x/>.@[*]; } > function () { <x/>.@[*]; } function () { <x/>[@[*]]; }

(In reply to comment #0) > This is a continuation of bug 349956. > > > function () { <x/>[@[*]]; } > function () { <x/>.@[*]; } I get the same result: js> function () { <x/>[@[*]]; } function () { <x/>.@[*]; } > > function () { <x/>.@[*]; } > function () { <x/>[@[*]]; } But I get the correct result here: js> function () { <x/>.@[*]; } function () { <x/>.@[*]; } I'm running with the patches for bug 349650 and bug 350531. /be

The JOF_XMLNAME format is as commented in jsopcode.h -- it must not apply to the JSOP_XMLNAME or derived (JSOP_(BIND|SET)XMLNAME) ops. Those ops are used only for unqualified (that is, not preceded by the binary . or .. operators) XML names, and so if used in [] after a left operand of [], must be decompiled using [], not by converting [] to dot. /be

Comment on attachment 237040 [details] [diff] [review] fix After staring at this long enough and after taking into (heavy) consideration the previous comment, I *think* this is correct.

Fixed on trunk. /be

Comment on attachment 237040 [details] [diff] [review] fix This fix is a pure E4X only decompilation fix, can't break anything else, and demonstrably helps. /be

Checking in regress-350226.js; /cvsroot/mozilla/js/tests/e4x/Regress/regress-350226.js,v <-- regress-350226.js initial revision: 1.1

Comment on attachment 237040 [details] [diff] [review] fix a=beltzner on behalf of 181drivers

To avoid conflicts, this patch should land after the patches for bug 350531 and its followups. /be

Fixed on the 1.8 branch. /be

verified fixed 1.9a1_2006090707 windows/mac*/linux

verified fixed 1.8 1.9 20060909 windows/mac*/linux