Fixed Typefeedback for Number Truncation

Fixed Typefeedback for Number Truncation

Additionally split BuildBinaryOperation into two parts for future usage
by a stub.

BUG=

[Sven Panne, 2013-08-01 09:21:09]

LGTM with nits.

https://codereview.chromium.org/21505002/diff/3001/src/hydrogen.cc
File src/hydrogen.cc (right):

https://codereview.chromium.org/21505002/diff/3001/src/hydrogen.cc#newcode7635
src/hydrogen.cc:7635: HValue* right,
Nit: Fix indentation of arguments.

https://codereview.chromium.org/21505002/diff/3001/src/hydrogen.cc#newcode7731
src/hydrogen.cc:7731: HValue* context) {
Hmmm, having 8 arguments is not very nice, but let's see how it will be used
elsewhere and perhaps refactor that. A TODO would be nice.

https://codereview.chromium.org/21505002/diff/3001/src/hydrogen.cc#newcode7736
src/hydrogen.cc:7736: if (op != Token::ADD ||
Brain explosion caused by heavy use of negation... o_O Pushing the negation
outwards, things are much clearer:

  if (!(op == Token::ADD &&
        (left_type->Maybe(Type::String()) ||
         right_type->Maybe(Type::String())))) {

Naming the expression would be even better:

  bool maybe_string_addition = ...
  if (!maybe_string_addition) {

This would even make the comment in the if body obsolete.

"Comments are a sign of our inability to express things in code." :-)

[oliv, 2013-08-01 09:53:43]

https://codereview.chromium.org/21505002/diff/3001/src/hydrogen.cc
File src/hydrogen.cc (right):

https://codereview.chromium.org/21505002/diff/3001/src/hydrogen.cc#newcode7731
src/hydrogen.cc:7731: HValue* context) {
On 2013/08/01 09:21:09, Sven Panne wrote:
> Hmmm, having 8 arguments is not very nice, but let's see how it will be used
> elsewhere and perhaps refactor that. A TODO would be nice.

well i know that the stub will have to pass exactly those args, since in the
case of stybs the type-feedback comes directly from the extra-ic-state and not
the Type system....

https://codereview.chromium.org/21505002/diff/3001/src/hydrogen.cc#newcode7736
src/hydrogen.cc:7736: if (op != Token::ADD ||
On 2013/08/01 09:21:09, Sven Panne wrote:
> Brain explosion caused by heavy use of negation... o_O Pushing the negation
> outwards, things are much clearer:
> 
>   if (!(op == Token::ADD &&
>         (left_type->Maybe(Type::String()) ||
>          right_type->Maybe(Type::String())))) {
> 
> Naming the expression would be even better:
> 
>   bool maybe_string_addition = ...
>   if (!maybe_string_addition) {
> 
> This would even make the comment in the if body obsolete.
> 
> "Comments are a sign of our inability to express things in code." :-)

yup, mechanical refactoring issue...