http://codereview.chromium.org/2130003/diff/1/2 File src/x64/codegen-x64.cc (right):
http://codereview.chromium.org/2130003/diff/1/2#newcode3452 src/x64/codegen-x64.cc:3452: if (FLAG_debug_code) __ AbortIfNotSmi(new_value.reg()); On 2010/05/17 13:02:35, Lasse Reichstein wrote:
__ is a macro. It may expand to more than one statement. You should
wrap the
then branch in brackets.
Done. http://codereview.chromium.org/2130003/diff/1/2#newcode5910 src/x64/codegen-x64.cc:5910: __ AbortIfNotSmi(left_side.reg()); On 2010/05/17 13:02:35, Lasse Reichstein wrote:
Do we ever use Abort* macros not guarded by FLAG_debug_code? If not, maybe we should put the flag test into the macro, or perhaps
make an
AssertSmi macro for it.
I think this is the case, but better do it as a separate change. It should also be changed in ia32 then. http://codereview.chromium.org/2130003/diff/1/3 File src/x64/macro-assembler-x64.cc (right): http://codereview.chromium.org/2130003/diff/1/3#newcode1739 src/x64/macro-assembler-x64.cc:1739: Assert(equal, "Operand not a number"); On 2010/05/17 13:02:35, Lasse Reichstein wrote:
What does "operand" signify here? Operand of what? How about "Unexpected non-number".
Yes, the message is not really meaningful, but I could not come up with sth. better. And it does not really matter so much. http://codereview.chromium.org/2130003/show -- v8-dev mailing list [email protected] http://groups.google.com/group/v8-dev
