Make invalid LHSs a parse-time (reference) error

Make invalid LHSs a parse-time (reference) error

This is required by the spec. It also prevents crashes resulting from the attempt to read type feedback for the RHS of an invalid assignment which full codegen never actually allocated info for.

To do: check properly in preparser already.

R=marja@chromium.org, mstarzinger@chromium.org
BUG=351658
LOG=Y

Committed: https://code.google.com/p/v8/source/detail?r=19976

[marja, 2014-03-17 08:48:45]

lgtm

[Michael Starzinger, 2014-03-17 09:28:17]

LGTM.

https://codereview.chromium.org/200473003/diff/1/test/cctest/test-parsing.cc
File test/cctest/test-parsing.cc (left):

https://codereview.chromium.org/200473003/diff/1/test/cctest/test-parsing.cc#...
test/cctest/test-parsing.cc:2221: "new foo++",
IIUC, this should throw an "invalid LHS" reference error now. Can we move it
into the ErrorsNewExpression test case? I assume the reason for removing it was
that parser and pre-parser don't agree (yet). In this case I would propose
leaving a TODO to add that test again once they do.

[rossberg, 2014-03-17 10:12:31]

https://codereview.chromium.org/200473003/diff/1/test/cctest/test-parsing.cc
File test/cctest/test-parsing.cc (left):

https://codereview.chromium.org/200473003/diff/1/test/cctest/test-parsing.cc#...
test/cctest/test-parsing.cc:2221: "new foo++",
On 2014/03/17 09:28:17, Michael Starzinger wrote:
> IIUC, this should throw an "invalid LHS" reference error now. Can we move it
> into the ErrorsNewExpression test case? I assume the reason for removing it
was
> that parser and pre-parser don't agree (yet). In this case I would propose
> leaving a TODO to add that test again once they do.

Done.

[rossberg, 2014-03-17 10:21:13]

Committed patchset #2 manually as r19976 (presubmit successful).