It thinks its a sequence of 'char' s in the to/fromCorba functions . Also the
idl was wrongly generated, mixing up the module scopes.
When I replace it with a vector<double>, it gets further (IDL is correct), but
I get the error:
[ 60%] Building CXX object transports/corba/CMakeFiles/foolib-transport-corba-
gnulinux.dir/__std__vector__double__.cpp.o
/home/kaltan/src/git/orocos-
toolchain/petproject/msg/transports/corba/__std__vector__double__.cpp:11:
error: redefinition of ‘struct RTT::corba::AnyConversion<std::vector
toolchain/install/include/rtt/transports/corba/CorbaConversion.hpp:381: error:
previous definition of ‘struct RTT::corba::AnyConversion<std::vector
gnulinux.dir/__std__vector__double__.cpp.o] Error 1
The only sane solution to me is to trim down RTT's CorbaConversion.hpp such
that it only contains the class definition and no specialisations. Or does
typegen rely on some of these specialisations ? I commented them out in
CorbaConversion.hpp, and the example compiled succesfully...
Still an impressive tool !!!
Peter
Attachment | Size |
---|---|
Foo.hpp | 216 bytes |
typegen: does not handle properly vector<string>, vector<double>
On Saturday 28 August 2010 22:44:41 Peter Soetens wrote:
> /home/kaltan/src/git/orocos-
> make[2]: *** [transports/corba/CMakeFiles/foolib-transport-corba-
> When I replace it with a vector<double>, it gets further (IDL is correct),
> but I get the error:
>
> [ 60%] Building CXX object
> transports/corba/CMakeFiles/foolib-transport-corba-
> gnulinux.dir/__std__vector__double__.cpp.o
> /home/kaltan/src/git/orocos-
> toolchain/petproject/msg/transports/corba/__std__vector__double__.cpp:11:
> error: redefinition of ‘struct
> RTT::corba::AnyConversion<std::vector
> toolchain/install/include/rtt/transports/corba/CorbaConversion.hpp:381:
> error: previous definition of ‘struct
> RTT::corba::AnyConversion<std::vector
> gnulinux.dir/__std__vector__double__.cpp.o] Error 1
>
> The only sane solution to me is to trim down RTT's CorbaConversion.hpp such
> that it only contains the class definition and no specialisations. Or does
> typegen rely on some of these specialisations ? I commented them out in
> CorbaConversion.hpp, and the example compiled succesfully...
I pushed this proposed fix to master already.
Peter
--
Orocos-Dev mailing list
Orocos-Dev [..] ...
http://lists.mech.kuleuven.be/mailman/listinfo/orocos-dev
typegen: does not handle properly vector<string>, vector<double>
>> The only sane solution to me is to trim down RTT's CorbaConversion.hpp such
>> that it only contains the class definition and no specialisations. Or does
>> typegen rely on some of these specialisations ? I commented them out in
>> CorbaConversion.hpp, and the example compiled succesfully...
IMO the proper solution is to finally remove vector<string> and
vector<double> from the RTT typekit. Having these there does not make
any sense.
And typegen does not expect any specialization to be there, so that fix
is fine from my POV.
--
Sylvain Joyeux (Dr. Ing.)
Researcher - Space and Security Robotics
DFKI Robotics Innovation Center
Bremen, Robert-Hooke-Straße 5, 28359 Bremen, Germany
Phone: +49 421 218-64136
Fax: +49 421 218-64150
Email: sylvain [dot] joyeux [..] ...
Weitere Informationen: http://www.dfki.de
--
Orocos-Dev mailing list
Orocos-Dev [..] ...
http://lists.mech.kuleuven.be/mailman/listinfo/orocos-dev
typegen: does not handle properly vector<string>, vector<double>
On Monday 30 August 2010 07:26:07 Sylvain Joyeux wrote:
> >> The only sane solution to me is to trim down RTT's CorbaConversion.hpp
> >> such that it only contains the class definition and no specialisations.
> >> Or does typegen rely on some of these specialisations ? I commented them
> >> out in CorbaConversion.hpp, and the example compiled succesfully...
>
> IMO the proper solution is to finally remove vector<string> and
> vector<double> from the RTT typekit. Having these there does not make
> any sense.
Right now they do, for ease-of-transitioning reasons. We could have a build
option that RTT only has the C/POD types.
>
> And typegen does not expect any specialization to be there, so that fix
> is fine from my POV.
>
Ok.
Peter
typegen: does not handle properly vector<string>, vector<double>
On 08/30/2010 02:06 PM, Peter Soetens wrote:
> On Monday 30 August 2010 07:26:07 Sylvain Joyeux wrote:
>>>> The only sane solution to me is to trim down RTT's CorbaConversion.hpp
>>>> such that it only contains the class definition and no specialisations.
>>>> Or does typegen rely on some of these specialisations ? I commented them
>>>> out in CorbaConversion.hpp, and the example compiled succesfully...
>>
>> IMO the proper solution is to finally remove vector<string> and
>> vector<double> from the RTT typekit. Having these there does not make
>> any sense.
>
> Right now they do, for ease-of-transitioning reasons. We could have a build
> option that RTT only has the C/POD types.
>
>>
>> And typegen does not expect any specialization to be there, so that fix
>> is fine from my POV.
>>
>
> Ok.
On a related matter, we should probably downgrade the "type XXX
registered multiple times" to INFO. This can happen, and I already have
a few cases where it is not a bad thing.
--
Sylvain Joyeux (Dr. Ing.)
Researcher - Space and Security Robotics
DFKI Robotics Innovation Center
Bremen, Robert-Hooke-Straße 5, 28359 Bremen, Germany
Phone: +49 421 218-64136
Fax: +49 421 218-64150
Email: sylvain [dot] joyeux [..] ...
Weitere Informationen: http://www.dfki.de
--
Orocos-Dev mailing list
Orocos-Dev [..] ...
http://lists.mech.kuleuven.be/mailman/listinfo/orocos-dev
typegen: does not handle properly vector<string>, vector<double>
On Tuesday 31 August 2010 04:12:53 Sylvain Joyeux wrote:
> On 08/30/2010 02:06 PM, Peter Soetens wrote:
> > On Monday 30 August 2010 07:26:07 Sylvain Joyeux wrote:
> >>>> The only sane solution to me is to trim down RTT's CorbaConversion.hpp
> >>>> such that it only contains the class definition and no
> >>>> specialisations. Or does typegen rely on some of these specialisations
> >>>> ? I commented them out in CorbaConversion.hpp, and the example
> >>>> compiled succesfully...
> >>
> >> IMO the proper solution is to finally remove vector<string> and
> >> vector<double> from the RTT typekit. Having these there does not make
> >> any sense.
> >
> > Right now they do, for ease-of-transitioning reasons. We could have a
> > build option that RTT only has the C/POD types.
> >
> >> And typegen does not expect any specialization to be there, so that fix
> >> is fine from my POV.
> >
> > Ok.
>
> On a related matter, we should probably downgrade the "type XXX
> registered multiple times" to INFO. This can happen, and I already have
> a few cases where it is not a bad thing.
Ok.
Peter
typegen: does not handle properly vector<string>, vector<double>
On Aug 30, 2010, at 01:26 , Sylvain Joyeux wrote:
>>> The only sane solution to me is to trim down RTT's CorbaConversion.hpp such
>>> that it only contains the class definition and no specialisations. Or does
>>> typegen rely on some of these specialisations ? I commented them out in
>>> CorbaConversion.hpp, and the example compiled succesfully...
>
> IMO the proper solution is to finally remove vector<string> and
> vector<double> from the RTT typekit. Having these there does not make
> any sense.
Uh, huh!? If you are thinking of totally removing these, maybe you should check with the community to see who is actually using these.
S