- Add a test case to test Source code generation of all the corelibs

- Add a test case to test Source code generation of all the corelibs
  (core, coreimpl, isolate, math and mirrors).

- Fix bugs in source code generation which were causing the above tests to
  fail.

(This should take care of issue 4647 and related bugs).

[hausner, 2012-09-07 22:12:29]

lgtm

https://chromiumcodereview.appspot.com/10911159/diff/1/vm/snapshot_test.cc
File vm/snapshot_test.cc (right):

https://chromiumcodereview.appspot.com/10911159/diff/1/vm/snapshot_test.cc#ne...
vm/snapshot_test.cc:730: static void GenerateSourceAndCheck(const Script&
script) {
You could pass in the private_key value that is used in the library where the
script came from. It seems you would not have to unmangle the names to compare
them.

[siva, 2012-09-09 04:57:29]

https://chromiumcodereview.appspot.com/10911159/diff/1/vm/snapshot_test.cc
File vm/snapshot_test.cc (right):

https://chromiumcodereview.appspot.com/10911159/diff/1/vm/snapshot_test.cc#ne...
vm/snapshot_test.cc:730: static void GenerateSourceAndCheck(const Script&
script) {
Good point.

Passed in private key and got rid of the unmangle code.

On 2012/09/07 22:12:29, hausner wrote:
> You could pass in the private_key value that is used in the library where the
> script came from. It seems you would not have to unmangle the names to compare
> them.

[Ivan Posva, 2012-09-09 10:24:17]

LGTM.

Thank you,
-Ivan

https://chromiumcodereview.appspot.com/10911159/diff/7002/vm/object.cc
File vm/object.cc (right):

https://chromiumcodereview.appspot.com/10911159/diff/7002/vm/object.cc#newcod...
vm/object.cc:5103: // TODO(4995): Temp solution for raw strings, this will break
I see two different options here to fix this current workaround: Either
introduce a token kRAW_STRING or prepend a token AT before the string literal
token. Obviously both of these would need some extra handling in the parser.

[Ivan Posva, 2012-09-09 10:31:58]

https://chromiumcodereview.appspot.com/10911159/diff/7002/vm/object.cc
File vm/object.cc (right):

https://chromiumcodereview.appspot.com/10911159/diff/7002/vm/object.cc#newcod...
vm/object.cc:5103: // TODO(4995): Temp solution for raw strings, this will break
On 2012/09/09 10:24:17, Ivan Posva wrote:
> I see two different options here to fix this current workaround: Either
> introduce a token kRAW_STRING or prepend a token AT before the string literal
> token. Obviously both of these would need some extra handling in the parser.

Of course I had another idea just after hitting send which does not need changes
to the parser: Preserve the fact that the string was a raw string in the literal
token.