Separate native and interpreted regexp by compile time flag, not runtime.

Separate native and interpreted regexp by compile time flag, not runtime.
Clean-up of RegExp code.

[Lasse Reichstein, 2009-07-06 12:52:18]

Review of suggested changes.

Erik: This breaks running of irregexp interpreter benchmarks. You'll need an
extra executable for that.

[bak, 2009-07-06 13:44:18]

LGTM,
  Lars

http://codereview.chromium.org/155085/diff/1/4
File src/jsregexp.cc (right):

http://codereview.chromium.org/155085/diff/1/4#newcode277
Line 277: // If it's a JS value, it's an error from a previous compilation.
Please change comment to reflect the test.
Note: JSValue is a subclass of JSObject.

http://codereview.chromium.org/155085/diff/1/4#newcode418
Line 418: Handle<FixedArray> array;
Two lines up: eliminate rc. Not really needed in the function.

http://codereview.chromium.org/155085/diff/1/4#newcode449
Line 449: if (!rc) return Factory::null_value();
if (res != RegExpMacroAssemblerIA32::SUCCESS) {
  return Factory::null_value();
}

http://codereview.chromium.org/155085/diff/1/4#newcode477
Line 477: rc = IrregexpInterpreter::Match(byte_codes,
if (!Irr...) return Factory::null_value();

http://codereview.chromium.org/155085/diff/1/5
File src/jsregexp.h (left):

http://codereview.chromium.org/155085/diff/1/5#oldcode40
Line 40: static inline bool UseNativeRegexp() {
You might want to keep: bool UsesNativeRegexp();

http://codereview.chromium.org/155085/diff/1/6
File src/objects-debug.cc (right):

http://codereview.chromium.org/155085/diff/1/6#newcode717
Line 717: #ifdef V8_NATIVE_REGEXP
It would probably be easier to use an accessor.

[Lasse Reichstein, 2009-07-07 08:09:36]

Moved the inline EnsureCompiledIrregexp definition to jsregexp.cc.
Abstracted out the code index computation.
Fixed xcode, visual studio and gyp build scripts to include V8_NATIVE_REGEXP
flag for ia32 builds.

http://codereview.chromium.org/155085/diff/1/4
File src/jsregexp.cc (right):

http://codereview.chromium.org/155085/diff/1/4#newcode277
Line 277: // If it's a JS value, it's an error from a previous compilation.
Done. Now uses JSObject about the value.

http://codereview.chromium.org/155085/diff/1/4#newcode418
Line 418: Handle<FixedArray> array;
Done.

http://codereview.chromium.org/155085/diff/1/4#newcode449
Line 449: if (!rc) return Factory::null_value();
Done.

http://codereview.chromium.org/155085/diff/1/4#newcode477
Line 477: rc = IrregexpInterpreter::Match(byte_codes,
On 2009/07/06 13:44:18, bak wrote:
> if (!Irr...) return Factory::null_value();

Done.

http://codereview.chromium.org/155085/diff/1/5
File src/jsregexp.h (left):

http://codereview.chromium.org/155085/diff/1/5#oldcode40
Line 40: static inline bool UseNativeRegexp() {
On 2009/07/06 13:44:18, bak wrote:
> You might want to keep: bool UsesNativeRegexp();
> 

Done.

http://codereview.chromium.org/155085/diff/1/6
File src/objects-debug.cc (right):

http://codereview.chromium.org/155085/diff/1/6#newcode717
Line 717: #ifdef V8_NATIVE_REGEXP
UsesNativeRegexp reinstated and used.