Refactor Parser to make it usable on a background thread.

Refactor Parser to make it usable on a background thread.

- Background Parsers cannot get the following data from Isolate (pass it to the
ctor instead): stack limit (background Parsers need a different stack limit),
UnicodeCache (background parsers need a separate UnicodeCache), hash seed
(Parser cannot access the Heap to get it). The Parser::Parse API won't change.

- Make the internalization phase (where Parser interacts with the heap) more
explicit. Previously, Parser was interacting with the heap here and there.

- Move HandleSourceURLComments out of DoParseProgram, so that background parsing
can use DoParseProgram too.

BUG=
R=rossberg@chromium.org

Committed: https://code.google.com/p/v8/source/detail?r=23600

[marja, 2014-09-02 08:06:49]

marja@chromium.org changed reviewers:
+ rossberg@chromium.org

[marja, 2014-09-02 08:06:49]

tentatively sending to rossberg@, please let me know if I should find somebody
else to review...

[rossberg, 2014-09-02 08:47:40]

lgtm

https://codereview.chromium.org/527763002/diff/20001/src/parser.h
File src/parser.h (right):

https://codereview.chromium.org/527763002/diff/20001/src/parser.h#newcode603
src/parser.h:603: Parser(CompilationInfo* info, uintptr_t stack_limit, uint32_t
hash_seed,
Would it be worth introducing a struct to collect these additional arguments?
E.g. ParseInfo or something?

[marja, 2014-09-02 09:48:40]

Thanks for review!

https://codereview.chromium.org/527763002/diff/20001/src/parser.h
File src/parser.h (right):

https://codereview.chromium.org/527763002/diff/20001/src/parser.h#newcode603
src/parser.h:603: Parser(CompilationInfo* info, uintptr_t stack_limit, uint32_t
hash_seed,
On 2014/09/02 08:47:40, rossberg wrote:
> Would it be worth introducing a struct to collect these additional arguments?
> E.g. ParseInfo or something?

Ok. Ultimately, we'd want is separate structs for things which are inputs to the
parse phase and for things which are outputs from the parse phase. That would
clarify things. Currently CompilationInfo has both...

Anyhow, I added ParserInfo here, to store these three things. Parser doesn't
need to store, because it only passes the information forward.

(I moved the AstValueFactory creation to Parser ctor, so we don't need to store
the hash seed any more. That would've been the right place for the code anyway.)

ParserInfo is passed the same way as CompilationInfo, i.e., the caller creates
it on the stack and passes a pointer (ownership not transferred).

[rossberg, 2014-09-02 10:01:07]

lgtm

[marja, 2014-09-02 11:36:33]

Committed patchset #5 (id:80001) manually as 23600 (presubmit successful).