Modify extension request IPC messages to pass a ListValue instead of a string.

Modify extension request IPC messages to pass a ListValue instead of a string.

This allows us to pass binary values through extension requests. I use this in
my next CL to pass SkBitmaps.

BUG=23269
TEST=no

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=28130

[Aaron Boodman, 2009-10-06 00:09:40]

Cool, my only real comment is about the weird situation with the 1-element list.

http://codereview.chromium.org/251093/diff/1/16
File chrome/browser/renderer_host/render_view_host.cc (right):

http://codereview.chromium.org/251093/diff/1/16#newcode1227
Line 1227: NOTREACHED();
Again with the << "message".

http://codereview.chromium.org/251093/diff/1/22
File chrome/renderer/extensions/extension_process_bindings.cc (right):

http://codereview.chromium.org/251093/diff/1/22#newcode367
Line 367: NOTREACHED();
Probably worth adding a little << "message" here. I hear that they get output to
the console in linux land.

http://codereview.chromium.org/251093/diff/1/22#newcode373
Line 373: args_holder.Append(json_args);
An extension request is fundamentally a list of (more than one) arguments. What
is going on here? I have this sick feeling we are having a list of one item,
which contains a list of items.

[Matt Perry, 2009-10-06 00:34:12]

http://codereview.chromium.org/251093/diff/1/16
File chrome/browser/renderer_host/render_view_host.cc (right):

http://codereview.chromium.org/251093/diff/1/16#newcode1227
Line 1227: NOTREACHED();
On 2009/10/06 00:09:40, Aaron Boodman wrote:
> Again with the << "message".

Done.

http://codereview.chromium.org/251093/diff/1/22
File chrome/renderer/extensions/extension_process_bindings.cc (right):

http://codereview.chromium.org/251093/diff/1/22#newcode367
Line 367: NOTREACHED();
On 2009/10/06 00:09:40, Aaron Boodman wrote:
> Probably worth adding a little << "message" here. I hear that they get output
to
> the console in linux land.

Done.

http://codereview.chromium.org/251093/diff/1/22#newcode373
Line 373: args_holder.Append(json_args);
On 2009/10/06 00:09:40, Aaron Boodman wrote:
> An extension request is fundamentally a list of (more than one) arguments.
What
> is going on here? I have this sick feeling we are having a list of one item,
> which contains a list of items.

I can't find an easy way around this. The Value we're sticking in here can be
anything.

[Aaron Boodman, 2009-10-06 00:51:15]

We would have to change the js side to always send a list (delete code that
special cases a single value), and change ExtensionFunction to make args_ a
ListValue, and then change all the ExtensionFunctions that are currently just
casting it to some value to get the first item out and cast that. It seems like
it should be pretty mechanical.

[Matt Perry, 2009-10-06 00:54:09]

On 2009/10/06 00:51:15, Aaron Boodman wrote:
> We would have to change the js side to always send a list (delete code that
> special cases a single value), and change ExtensionFunction to make args_ a
> ListValue, and then change all the ExtensionFunctions that are currently just
> casting it to some value to get the first item out and cast that. It seems
like
> it should be pretty mechanical.

Yeah. It just seems like a lot of grunt work for little gain.

[Aaron Boodman, 2009-10-06 01:02:24]

Ok fine.

lgtm

On Mon, Oct 5, 2009 at 5:54 PM,  <mpcomplete@chromium.org> wrote:
> On 2009/10/06 00:51:15, Aaron Boodman wrote:
>>
>> We would have to change the js side to always send a list (delete code
>> that
>> special cases a single value), and change ExtensionFunction to make args_
>> a
>> ListValue, and then change all the ExtensionFunctions that are currently
>> just
>> casting it to some value to get the first item out and cast that. It seems
>
> like
>>
>> it should be pretty mechanical.
>
> Yeah. It just seems like a lot of grunt work for little gain.
>
> http://codereview.chromium.org/251093
>