Pass pending exception to the message listener.

Pass pending exception to the message listener.


BUG=

Committed: https://code.google.com/p/v8/source/detail?r=12768

[Yang, 2012-10-01 11:54:28]

This is to make the message listener be able to access the error object that is
causing the thrown exception.

Even though this does not change the API syntax, it changes its semantics, as
described in v8.h. We would be using the data argument for the preset value if
available, or the pending exception if former is not.

Alternatives that may be cleaner are:
- Changes to the API syntax, adding one more argument to MessageCallback to pass
the exception.
- Piggyback the Error object on the Message object by adding a new field to the
latter.

[Yang, 2012-10-01 13:14:27]

On 2012/10/01 11:54:28, Yang wrote:
> This is to make the message listener be able to access the error object that
is
> causing the thrown exception.
> 
> Even though this does not change the API syntax, it changes its semantics, as
> described in v8.h. We would be using the data argument for the preset value if
> available, or the pending exception if former is not.
> 
> Alternatives that may be cleaner are:
> - Changes to the API syntax, adding one more argument to MessageCallback to
pass
> the exception.
> - Piggyback the Error object on the Message object by adding a new field to
the
> latter.

I haven't found any use for the optional second argument of AddMessageListener
(either in Chromium or Node.js), so I assume that it's an ancient relics of a
time long forgotten. I'm removing the second argument entirely. This has the
advantage that:
- We can be sure of the meaning of the second argument in the MessageCallback.
If some embedder is indeed still using it to get the second argument for
AddMessageListener, he will notice simply because AddMessageListener stops
working that way.
- We don't have to change the definition for MessageCallback, as the signature
remains the same. This saves us a lot of hassle sync'ing the change with WebKit.

[ulan, 2012-10-01 13:22:50]

LGTM if it doesn't break webkit layout tests.

[Mike West, 2012-10-17 12:27:47]

On 2012/10/01 13:22:50, ulan wrote:
> LGTM if it doesn't break webkit layout tests.

Yang? Would you mind rebasing this on top of ToT? It won't let me upload a new
patch to this review. :(

[Yang, 2012-10-17 12:40:17]

On 2012/10/17 12:27:47, Mike West (chromium) wrote:
> On 2012/10/01 13:22:50, ulan wrote:
> > LGTM if it doesn't break webkit layout tests.
> 
> Yang? Would you mind rebasing this on top of ToT? It won't let me upload a new
> patch to this review. :(

just rebased. I guess trybots won't work though since this refers to the v8
repository and not Chromium.