On Thursday 17 February 2011 16:31:51 Peter Soetens wrote:
> On Thursday 17 February 2011 15:51:56 Ruben Smits wrote:
> > On Thursday 17 February 2011 14:34:46 Ruben Smits wrote:
> > > Hi,
> > >
> > > I'm having some trouble connecting the ctaskbrowser with a remote
> > > component.
> > >
> > > When I launch the ctaskbrowser I get the following errors:
> > >
> > > 0.080 [ Warning][TypekitRepository::Import] Overriding TypeInfo for
> > > 'int' with 'int32'.
> > > 0.080 [ Warning][TypekitRepository::Import] Overriding TypeInfo for
> > > 'uint' with 'uint32'.
> > > 0.080 [ Warning][TypekitRepository::Import] Overriding TypeInfo for
> > > 'float' with 'float32'.
> > > 0.080 [ Warning][TypekitRepository::Import] Overriding TypeInfo for
> > > 'double' with 'float64'.
> > > 0.080 [ Warning][TypekitRepository::Import] Overriding TypeInfo for
> > > 'array' with 'float64[]'.
> > > 0.080 [ Warning][TypekitRepository::Import] Overriding TypeInfo for
> > > 'string' with 'string'.
> > > 0.080 [ Warning][TypekitRepository::Import] Registering Type name
> > > 'string' twice to the Orocos Type System: replacing.
> > > 0.085 [ ERROR ][TaskContextProxy] Looking up Attribute int32: type
> > > not
> > > known. Check your RTT_COMPONENT_PATH.
> > >
> > > When I try to call an operation with the following (c++)signature:
> > >
> > > bool SoemEL2xxx::switchOn(unsigned int n)
> > >
> > > I get the following errors:
> > >
> > > ctaskbrowser side:
> > >
> > > Master.Slave_1002.switchOn(0)
> > >
> > > 56.727 [ ERROR ][CorbaFallBackProtocol] Could not send data of type
> > > 'int32'
> > >
> > > : data type not known to CORBA Transport.
> > >
> > > Wrong number of arguments in call of function "this.switchOn":
> > > expected
> > > 1, received 0.
> > >
> > > remote side:
> > > 60.989 [ ERROR ][Master] Registered transport for type uint32 could
> > > not create data source from Any (argument 1): calling operation
> > > 'switchOn' will fail.
> > >
> > > I guess this is due to the fact that there are no CORBA transports
> > > defined for these basic types, but how could I create these?
> >
> > I created a tranportkit using the CorbaTemplateProtocol:
> > [...]
> >
> > bool registerTransport(std::string name, types::TypeInfo*
> > ti)
> > {
> >
> > assert( name == ti->getTypeName() );
> > if ( name == "int32")
> >
> > return
> > ti->addProtocol(ORO_CORBA_PROTOCOL_ID, new
> >
> > corba::CorbaTemplateProtocol<int32_t>());
> > if ( name == "uint32")
> >
> > return
> > ti->addProtocol(ORO_CORBA_PROTOCOL_ID, new
> >
> > corba::CorbaTemplateProtocol<uint32_t>());
> > [...]
>
> This won't work always. The big problem with 2.2 code is that it doesn't
> know how to alias. For example, the assert has to go certainly. Cfr
> Charles' mail. The rest is ok.
>
> > The transport is loaded:
> >
> > 0.148 [ Info ][TypekitRepository::Import] Registered new 'CORBA'
> > transport for int32
> > 0.148 [ Info ][TypekitRepository::Import] Registered new 'CORBA'
> > transport for uint32
> >
> > And the error about the transport dissapears but the ctaskbrowser is
> > still complaining about the arguments;
> >
> > ctaskbrowser side:
> > Master [R]> Slave_1002.switchOn(0)
> > Wrong number of arguments in call of function "this.switchOn": expected
> > 1, received 0.
> >
> > remote side:
> > 168.893 [ ERROR ][Logger] Registered transport for type uint32 could
> > not
> > create data source from Any (argument 1): calling operation 'switchOn'
> > will fail.
>
> This probably means that the received arguments (which are any objects)
> contain a wrong type, or at least a type it doesn't expect.
>
> > Am I still missing something or cann't I use the CorbaTemplateProtocol
> > for these basic types?
>
> The CorbaTemplateProtocol class should be fine. The main thing to avoid is
> to mix RTT types and ROS types in 2.2. I think your approach of making a
> new Corba transport typekit for 'ros-style' primitive C types is the best
> thing to do.

Well, apparently this approach of making a Corba transport typekit deos not
work, as I already created this typekit but it is still not working as shown
in my previous email :(

--Ruben

> Peter