Fixed compilation time measurement, restructure some code.

Fixed compilation time measurement, restructure some code. Do not use slow ToFullyQualifiedCString in production backend (saves 10% in backend time with compile-all).
Committed: https://code.google.com/p/dart/source/detail?r=7435

[regis, 2012-05-08 21:26:41]

LGTM

https://chromiumcodereview.appspot.com/10389023/diff/6001/runtime/vm/code_gen...
File runtime/vm/code_generator_ia32.h (right):

https://chromiumcodereview.appspot.com/10389023/diff/6001/runtime/vm/code_gen...
runtime/vm/code_generator_ia32.h:80: virtual void InitGenerator();
Add comment explaining why you factorize this initialization out of the
constructor.

[hausner, 2012-05-08 21:41:01]

LGTM with comments

https://chromiumcodereview.appspot.com/10389023/diff/6001/runtime/vm/object.cc
File runtime/vm/object.cc (right):

https://chromiumcodereview.appspot.com/10389023/diff/6001/runtime/vm/object.c...
runtime/vm/object.cc:3656: const char* library_name = 
String::Handle(library.url()).ToCString();
Extra space to balance the universe?

https://chromiumcodereview.appspot.com/10389023/diff/6001/runtime/vm/object.c...
runtime/vm/object.cc:6072: 
Remove extra line?

https://chromiumcodereview.appspot.com/10389023/diff/6001/runtime/vm/object.c...
runtime/vm/object.cc:6075: RawCode* Code::FinalizeCode(const char* name,
Maybe an assert in this function that name is not "" when gdb or pprof info is
generated?

https://chromiumcodereview.appspot.com/10389023/diff/6001/runtime/vm/object.c...
runtime/vm/object.cc:6165: 
Save the extra empty line for a rainy day.

https://chromiumcodereview.appspot.com/10389023/diff/6001/runtime/vm/timer.h
File runtime/vm/timer.h (left):

https://chromiumcodereview.appspot.com/10389023/diff/6001/runtime/vm/timer.h#...
runtime/vm/timer.h:41: // Get total cummulative elapsed time in micros and reset
the counters.
Please adjust the comment. I think this was originally Siva's code, so me have
an answer as to why the Reset was here.

[srdjan, 2012-05-08 21:43:17]

https://chromiumcodereview.appspot.com/10389023/diff/6001/runtime/vm/code_gen...
File runtime/vm/code_generator_ia32.h (right):

https://chromiumcodereview.appspot.com/10389023/diff/6001/runtime/vm/code_gen...
runtime/vm/code_generator_ia32.h:80: virtual void InitGenerator();
On 2012/05/08 21:26:41, regis wrote:
> Add comment explaining why you factorize this initialization out of the
> constructor.

  // Constructor is lighweight, major initialization work should occur here.
  // This makes it easier to measure time spent in the code generator.

[srdjan, 2012-05-08 21:51:31]

https://chromiumcodereview.appspot.com/10389023/diff/6001/runtime/vm/object.cc
File runtime/vm/object.cc (right):

https://chromiumcodereview.appspot.com/10389023/diff/6001/runtime/vm/object.c...
runtime/vm/object.cc:3656: const char* library_name = 
String::Handle(library.url()).ToCString();
On 2012/05/08 21:41:01, hausner wrote:
> Extra space to balance the universe?

Rebalanced the universe.

https://chromiumcodereview.appspot.com/10389023/diff/6001/runtime/vm/object.c...
runtime/vm/object.cc:6072: 
On 2012/05/08 21:41:01, hausner wrote:
> Remove extra line?

Done.

https://chromiumcodereview.appspot.com/10389023/diff/6001/runtime/vm/object.c...
runtime/vm/object.cc:6075: RawCode* Code::FinalizeCode(const char* name,
On 2012/05/08 21:41:01, hausner wrote:
> Maybe an assert in this function that name is not "" when gdb or pprof info is
> generated?

Done.

https://chromiumcodereview.appspot.com/10389023/diff/6001/runtime/vm/object.c...
runtime/vm/object.cc:6165: 
On 2012/05/08 21:41:01, hausner wrote:
> Save the extra empty line for a rainy day.

Done.

https://chromiumcodereview.appspot.com/10389023/diff/6001/runtime/vm/timer.h
File runtime/vm/timer.h (left):

https://chromiumcodereview.appspot.com/10389023/diff/6001/runtime/vm/timer.h#...
runtime/vm/timer.h:41: // Get total cummulative elapsed time in micros and reset
the counters.
On 2012/05/08 21:41:01, hausner wrote:
> Please adjust the comment. I think this was originally Siva's code, so me have
> an answer as to why the Reset was here. 

Done.

[siva, 2012-05-08 22:08:45]

LGTM

https://chromiumcodereview.appspot.com/10389023/diff/12004/runtime/vm/compile...
File runtime/vm/compiler.cc (right):

https://chromiumcodereview.appspot.com/10389023/diff/12004/runtime/vm/compile...
runtime/vm/compiler.cc:191: Code::Handle(Code::FinalizeCodeForFunction(function,
&assembler));
Why not have FinalizeCode and FinalizeStubCode one takes the function and the
other name

https://chromiumcodereview.appspot.com/10389023/diff/12004/runtime/vm/compile...
runtime/vm/compiler.cc:218: }
You may get a compilation error on the build bot because
we are relying on the initialized value of 'is_compiled' here.
Maybe just set is_compiled = false; and get rid of the
initialization above.

[srdjan, 2012-05-08 23:54:31]

https://chromiumcodereview.appspot.com/10389023/diff/12004/runtime/vm/compile...
File runtime/vm/compiler.cc (right):

https://chromiumcodereview.appspot.com/10389023/diff/12004/runtime/vm/compile...
runtime/vm/compiler.cc:218: }
On 2012/05/08 22:08:45, asiva wrote:
> You may get a compilation error on the build bot because
> we are relying on the initialized value of 'is_compiled' here.
> Maybe just set is_compiled = false; and get rid of the
> initialization above.

Done.