Make PreParser track valid left hand sides.

Make PreParser track valid left hand sides.

Notes:
- This makes PreParser produce invalid_lhs_in_assignment and
invalid_lhs_in_prefix_op. Other errors will follow as the corresponding funcs
move to ParserBase.
- PreParserExpression::IsStrictFunction and StrictFunction() are not needed any
more -> removed them.

R=rossberg@chromium.org
BUG=

Committed: https://code.google.com/p/v8/source/detail?r=20125

[marja, 2014-03-19 15:07:49]

rossberg, ptal

[rossberg, 2014-03-19 15:29:08]

LGTM modulo comments

Nice, this is simpler than I thought it'd be.

https://codereview.chromium.org/196343033/diff/1/test/cctest/test-parsing.cc
File test/cctest/test-parsing.cc (right):

https://codereview.chromium.org/196343033/diff/1/test/cctest/test-parsing.cc#...
test/cctest/test-parsing.cc:2336: // "new foo ++",
Can we activate this one now?

https://codereview.chromium.org/196343033/diff/1/test/cctest/test-parsing.cc#...
test/cctest/test-parsing.cc:2592: // TODO(marja): This doesn't work yet.
Hm, why?

[marja, 2014-03-19 15:41:14]

Thanks for review!

I was also expecting this to be more difficult, but reading
Expression::IsValidLeftHandSide, it seems that the only rules are:
- property expressions are valid left hand sides
- variables (i.e., identifiers) are valid left hand sides

There is one place where variables which are not valid left hand sides are
created, in Scope::Initialize, but that's for some "convenience variable" and
shouldn't be relevant.

(Comments NOOP, will commit the latest patch)

https://codereview.chromium.org/196343033/diff/1/test/cctest/test-parsing.cc
File test/cctest/test-parsing.cc (right):

https://codereview.chromium.org/196343033/diff/1/test/cctest/test-parsing.cc#...
test/cctest/test-parsing.cc:2336: // "new foo ++",
On 2014/03/19 15:29:08, rossberg wrote:
> Can we activate this one now?

Not yet; this will start working when ParseMemberWithNewPrefixesExpression is
moved to Base.

https://codereview.chromium.org/196343033/diff/1/test/cctest/test-parsing.cc#...
test/cctest/test-parsing.cc:2592: // TODO(marja): This doesn't work yet.
On 2014/03/19 15:29:08, rossberg wrote:
> Hm, why?

Because the check is in ParsePostfixExpression which is not yet unified, and
PreParser::ParsePostFixExpression contains no check. In the working cases, the
checks are in funcs which are already moved to Base.

The division point here is that this CL just adds this tracking information, so
funcs which are in Base already get to use it. But I didn't want to add all the
checks in this CL.

I'll unify ParsePostFixExpression in the next CL and enable it. (There are a
couple of other places in parser where this check is done, and they'll also get
unified eventually.)

[rossberg, 2014-03-19 15:55:31]

On 2014/03/19 15:41:14, marja wrote:
> Thanks for review!
> 
> I was also expecting this to be more difficult, but reading
> Expression::IsValidLeftHandSide, it seems that the only rules are:
> - property expressions are valid left hand sides
> - variables (i.e., identifiers) are valid left hand sides
> 
> There is one place where variables which are not valid left hand sides are
> created, in Scope::Initialize, but that's for some "convenience variable" and
> shouldn't be relevant.
> 
> (Comments NOOP, will commit the latest patch)
> 
> https://codereview.chromium.org/196343033/diff/1/test/cctest/test-parsing.cc
> File test/cctest/test-parsing.cc (right):
> 
>
https://codereview.chromium.org/196343033/diff/1/test/cctest/test-parsing.cc#...
> test/cctest/test-parsing.cc:2336: // "new foo ++",
> On 2014/03/19 15:29:08, rossberg wrote:
> > Can we activate this one now?
> 
> Not yet; this will start working when ParseMemberWithNewPrefixesExpression is
> moved to Base.
> 
>
https://codereview.chromium.org/196343033/diff/1/test/cctest/test-parsing.cc#...
> test/cctest/test-parsing.cc:2592: // TODO(marja): This doesn't work yet.
> On 2014/03/19 15:29:08, rossberg wrote:
> > Hm, why?
> 
> Because the check is in ParsePostfixExpression which is not yet unified, and
> PreParser::ParsePostFixExpression contains no check. In the working cases, the
> checks are in funcs which are already moved to Base.
> 
> The division point here is that this CL just adds this tracking information,
so
> funcs which are in Base already get to use it. But I didn't want to add all
the
> checks in this CL.
> 
> I'll unify ParsePostFixExpression in the next CL and enable it. (There are a
> couple of other places in parser where this check is done, and they'll also
get
> unified eventually.)

I see. Thanks.

[marja, 2014-03-19 17:13:54]

There was one fix needed, in PreParser::ParseMemberWithNewPrefixesExpression it
didn't transmit the lefthandsideness of the expression properly.

You can see the fix & the tests I added for it as diff between patch sets 1 and
2.

rossberg@, if you want to have a look at the latest version, ptal. Otherwise
I'll just commit (tomorrow).

[rossberg, 2014-03-19 17:15:55]

lgtm

[marja, 2014-03-20 13:18:26]

Committed patchset #3 manually as r20125 (presubmit successful).