Properly treat exceptions thrown while compiling.

Properly treat exceptions thrown while compiling.

BUG=v8:1132
TEST=test/mjsunit/regress/regress-1132.js

Committed: http://code.google.com/p/v8/source/detail?r=6754

[antonm, 2011-02-11 11:14:52]

Kevin, may you have a look, please?

I am not best versed in this part of the system, so there might be better way to
attack the problem---exception thrown in ParseLazy.  For example, it might be
not the best idea to disable optimization, although it at the very least
correct.

[Kevin Millikin (Chromium), 2011-02-11 11:43:18]

Change LGTM.  Go ahead and commit everything as is, and I'll see what we can do
about the regression test.

[antonm, 2011-02-11 12:49:18]

Kevin, may you have another quick look?

I've added flag to restrict v8 stack region size and that makes test pass ways
faster.  Running test on other platforms, fingers-crossedly.

yours,
anton.

On 2011/02/11 11:43:18, Kevin Millikin wrote:
> Change LGTM.  Go ahead and commit everything as is, and I'll see what we can
do
> about the regression test.

[Kevin Millikin (Chromium), 2011-02-11 12:55:47]

Still LGTM.

http://codereview.chromium.org/6487021/diff/4001/src/flag-definitions.h
File src/flag-definitions.h (right):

http://codereview.chromium.org/6487021/diff/4001/src/flag-definitions.h#newco...
src/flag-definitions.h:235: DEFINE_int(stack_size_kb, kPointerSize * 128,
For the heap limits, the flag is something like max_new_space_size and the
string help comment gives the units.

You could do that here for consistency, but I don't care too much either way.

[antonm, 2011-02-11 14:20:31]

Thanks a lot for review, Kevin.

I also raised the constant slightly to allow test to run under x64 and disabled
it for ARM.

Submitting

http://codereview.chromium.org/6487021/diff/4001/src/flag-definitions.h
File src/flag-definitions.h (right):

http://codereview.chromium.org/6487021/diff/4001/src/flag-definitions.h#newco...
src/flag-definitions.h:235: DEFINE_int(stack_size_kb, kPointerSize * 128,
On 2011/02/11 12:55:47, Kevin Millikin wrote:
> For the heap limits, the flag is something like max_new_space_size and the
> string help comment gives the units.
> 
> You could do that here for consistency, but I don't care too much either way.

Unified with max_new_space_size.