dbus: equal signal and method names

[GoogleCodeExporter]

So I just wanted to write a gadget that uses some DBus controls
of my application and i had the following issue:

1. provide a dbus interface with a method called "foo" and a signal called
"foo"
2. use a gadget to connect to the interface
3. try to assign a function to dbusobj.foo

you get: "ReferenceError: Failed to set native property message (may be
readonly)."

After a little tracing, it seems to me, that I've figured out what goes wrong.
As I understood the code, the dynamic getter of the DBus scriptable
determines the prototype of the property (if it's a signal, use a slot, if
it's a method, it's a scriptable). But to assign signals, the DynamicSetter
is used, which handles signals first (instead of the getter which handles
methods first). I built some checkpoints in the code:

T 22:45.559: main.js:89: ScriptableDBusObject::DynamicGetter(message)
T 22:45.559: main.js:89: well, trying to get method message.
T 22:45.559: main.js:89: Oh hey, here in native_js_wrapper,
SetPropertyByName (message).
T 22:45.559: main.js:89: ScriptableDBusObject::DynamicGetter(message)
T 22:45.559: main.js:89: well, trying to get method message.
T 22:45.559: main.js:89: prototype of message is 7.
T 22:45.560: main.js:89: ScriptableDBusObject::DynamicSetter(message)
T 22:45.560: main.js:89: searching for message.
T 22:45.560: main.js:89: got SCRIPTABLE:0xfb3be0(CLASS_ID=65f4d888b7b749ed)
in first place.
T 22:45.560: main.js:89: Signal property expects a slot not 7.
E 22:45.560: main.js:89: ReferenceError: Failed to set native property
message (may be readonly).

To summarize the log:
DynamicSetter is called to assign the function to the signal "message"
DynamicSetter finds the signal "message"
DynamicGetter is called to get the prototype.
DynamicGetter finds the method "message", prototype Variant::TYPE_SCRIPTABLE
DynamicSetter fails at value.type() == Variant::TYPE_SLOT
Assignment (false) = function() {} fails.

Original issue reported on code.google.com by pro...@gmail.com on 7 Aug 2009 at 7:45

[GoogleCodeExporter]

Your remote dbus object must support Introspectable interface, otherwise signal 
won't
work correctly.

Original comment by james...@gmail.com on 8 Aug 2009 at 2:23

[GoogleCodeExporter]

Well, i think it does. See the attached dbus-monitor log, captured while 
connecting
the widget to the bus.

Original comment by pro...@gmail.com on 8 Aug 2009 at 1:19

Attachments:

• dbus_log

[GoogleCodeExporter]

Then, which signal did you try to connect?

Original comment by james...@gmail.com on 8 Aug 2009 at 1:30

[GoogleCodeExporter]

message. There's a signal message and a method message which leads to type 
mismatch,
i think.

Original comment by pro...@gmail.com on 8 Aug 2009 at 1:32

[GoogleCodeExporter]

You can uncomment line 38 of ggadget/dbus/dbus_proxy.cc to get more verbose log 
message.

Original comment by james...@gmail.com on 8 Aug 2009 at 1:33

[GoogleCodeExporter]

That's the problem. they shouldn't use the same name, otherwise there is no way 
to
distinguish which you want when accessing the name from JavaScript.

Original comment by james...@gmail.com on 8 Aug 2009 at 1:59

[GoogleCodeExporter]

Did so. 

Original comment by pro...@gmail.com on 8 Aug 2009 at 2:00

Attachments:

• gadget_log

[GoogleCodeExporter]

You're right. With this architecture you use (prototype determination by dynamic
getter) it's not possible. Maybe you could implement a method to return a 
signal slot
for cases like this. It's not that unlikely that there's a signal and a method 
with
the same name, i think.

Original comment by pro...@gmail.com on 8 Aug 2009 at 2:02

[GoogleCodeExporter]

You can try to modify ScriptableDBusObject::DynamicGetter() to try signal before
method. Then you can access the signal. You may still call the method with
$callMethod method.

If it works for you, I can modify the trunk code as it.

Original comment by james...@gmail.com on 8 Aug 2009 at 2:20

[GoogleCodeExporter]

I did so before. Yes it works.

Original comment by pro...@gmail.com on 8 Aug 2009 at 2:27

[GoogleCodeExporter]

Ok, I'll modify trunk.

Original comment by james...@gmail.com on 8 Aug 2009 at 2:29

[GoogleCodeExporter]

Ok. Thank you.

Original comment by pro...@gmail.com on 8 Aug 2009 at 3:08

[GoogleCodeExporter]

Original comment by james...@gmail.com on 10 Aug 2009 at 9:53

• Changed state: Fixed