Update test262 expectations concerning 64-bit precision double.

test/test262/test262.status

Index: test/test262/test262.status
diff --git a/test/test262/test262.status b/test/test262/test262.status
index 16a478ba119bd0c96824a3d85972f6937a4d57d6..c912dcebb77ee72c96f65e7dbb95ffd0cd396a77 100644
--- a/test/test262/test262.status
+++ b/test/test262/test262.status
@@ -70,8 +70,8 @@ S7.8.4_A7.2_T6: FAIL_OK
 
 # Linux and Mac defaults to extended 80 bit floating point format in the FPU.
 # We follow the other major JS engines by keeping this default.
-S8.5_A2.2: PASS, FAIL if $system == linux, FAIL if $system == macos
-S8.5_A2.1: PASS, FAIL if $system == linux, FAIL if $system == macos
+S8.5_A2.2: PASS, FAIL if ($system == linux && ($arch == ia32 || $arch == arm))
+S8.5_A2.1: PASS, FAIL if ($system == linux && ($arch == ia32 || $arch == arm))

[Michael Starzinger, 2012/02/29 16:33:16]

I checked whether this syntax works, unfortunately it doesn't, it just marks the
test case as being flakey. But the following syntax does the trick:

[ $system == linux && ($arch == ia32 || $arch == arm) ]
S8.5_A2.1: FAIL_OK
S8.5_A2.2: FAIL_OK

Also I suspect this to fail on MIPS as well, so we might want to add that to the
list as well.

 
 ############################ SKIPPED TESTS #############################