Patch updated and ready for review.
Sorry for uploading so many patches, it seems my presubmit check wasn't
running
the linter :).
http://codereview.chromium.org/1694011/diff/35007/38002
File include/v8.h (right):
http://codereview.chromium.org/1694011/diff/35007/38002#newcode808
include/v8.h:808: static const int kNotFound;
On 2010/05/04 09:59:18, Michail Naganov wrote:
On 2010/05/03 19:16:02, jaimeyap wrote:
> On 2010/05/03 07:59:21, Søren Gjesse wrote:
> > In the profiler API this is called
CpuProfileNode::kNoLineNumberInfo and the
> > class Message can also return a line number. We should only have
one
constant
> > for this, and I suggest you place it in the class Message and call
it
> > kNoLineNumberInfo. Then update the comment in class Message and
make sure a
> > consistent value is returned (either 0 or -1 - I am not sure what
message
> does).
>
> I think Message returns -1. But CpuProfileNode and this new
StackTrace API
> return 0.
>
> Given that the indexes are all 1-based, I think returning 0 makes
most sense.
I
> changed them all to use Message::kNoLineNumberOffset, which ==0.
>
> All unit tests pass, so if anyone was depending on this being -1
then they may
> be broken :(.
>
As for the profiler API, I chose 0 because WebKit uses unsigned ints
for
reporting line numbers, so -1 was turning into 2^32 - 1, that isn't
acceptable.
I standardized on 0 as the notfound return value for Messages,
StackTraces and CpuProfiler.
http://codereview.chromium.org/1694011/show
--
v8-dev mailing list
[email protected]
http://groups.google.com/group/v8-dev
