[ros-users] Message proxy?

[William Woodall]

Another existing option is the multimaster stack (different from the
multimaster_experimental that rosproxy is part of).

http://ros.org/wiki/multimaster

--

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
William Woodall
Graduate Software Engineering
Auburn University
w at auburn.edu
wjwwood at gmail.com
williamjwoodall.com
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~



On Mon, Jul 2, 2012 at 4:25 PM, Austin Hendrix <legotown at aol.com> wrote:

> I would take a look at rosproxy: http://ros.org/wiki/rosproxy
>
> The serialization in ROS should be cross-platform/endianness.
>
> -Austin
>
>
> On 07/02/2012 12:46 PM, William Woodall wrote:
>
>> Sophisticated as this question is, I think it is still serviceable via
>> http://answers.ros.org/**questions/ <http://answers.ros.org/questions/>
>>
>> For part 1):
>>
>> Not sure, potentially if the code that deserializes was generated for an
>> ARM target.  Maybe some else can comment on this.
>>
>> For part 2):
>>
>> The message type '<pkg>/<MessageName>', the actual message definition,
>> and the md5sum of the message are all part of the data being sent in the
>> serialized data. See:
>>
>> http://www.ros.org/wiki/ROS/**TCPROS <http://www.ros.org/wiki/ROS/TCPROS>
>> http://www.ros.org/wiki/ROS/**Connection%20Header<http://www.ros.org/wiki/ROS/Connection%20Header>
>>
>> This is how, from what I understand, rosbag can record and republish ros
>> topic data with out linking to any message type headers, because it never
>> actually deserializes anything, similarly your proxies can forward the data
>> in its serialized form without ever knowing the type and ros nodes on the
>> other side can still determine the type and then deserialize the data.
>>
>> --
>>
>> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~**~
>> William Woodall
>> Graduate Software Engineering
>> Auburn University
>> w at auburn.edu <mailto:w at auburn.edu>
>> wjwwood at gmail.com <mailto:wjwwood at gmail.com>
>> williamjwoodall.com <http://williamjwoodall.com>
>> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~**~
>>
>>
>>
>>
>> On Mon, Jul 2, 2012 at 2:20 PM, Karan, Cem F CIV (US) <
>> cem.f.karan.civ at mail.mil <mailto:cem.f.karan.civ at mail.**mil<cem.f.karan.civ at mail.mil>>>
>> wrote:
>>
>>     Hi all, I'm trying to pass data between multiple robots in a manner
>>     similar to what multimaster (http://ros.org/wiki/**multimaster<http://ros.org/wiki/multimaster>)
>> is/was
>>     supposed to do, but after some work, I've realized that what I'm
>>     really
>>     trying to do is proxy messages from one robot to another.  I can
>>     do this
>>     by grabbing a message and asking it for its serialized
>>     representation, and
>>     then rsyncing that across the network, but I realized that that
>>     leaves 2
>>     potential problems:
>>
>>     1) I don't know how portable the serialization format really is.
>>      Can I
>>     take a message that was serialized on a little endian 32 bit
>>     machine (x86)
>>     and forward it to a big endian 16 bit machine (ARM), and have the
>>     data be
>>     correctly serialized/deserialized?  I know I can do this via google
>>     protocol buffers (https://developers.google.**com/protocol-buffers/<https://developers.google.com/protocol-buffers/>)
>> or
>>     MessagePack (http://msgpack.org/), but I'm not too sure if I can
>>     do this
>>     with serialized message format that ROS uses.
>>
>>     2) Is the message type encoded into the serialized data?  It doesn't
>>     appear so, but I just wanted to make sure.  This affects what kind of
>>     proxy I can make; if the message type is encoded in the message,
>>     then the
>>     proxy can be oblivious to messages, and will simply forward the data
>>     along.  The receiver can likewise be oblivious, putting the
>>     message 'on
>>     the bus' and letting all listeners decode what they want.
>>      Conversely, if
>>     the messages don't encode their own type, then the proxies will
>>     need to
>>     know every message type that they encode; otherwise, the receivers
>>     won't
>>     be able to put the right message type 'on the bus'.
>>
>>     Any help/pointers would be appreciated.
>>
>>     Thanks,
>>     Cem Karan
>>
>>     ______________________________**_________________
>>     ros-users mailing list
>>     ros-users at code.ros.org <mailto:ros-users at code.ros.org**>
>>     https://code.ros.org/mailman/**listinfo/ros-users<https://code.ros.org/mailman/listinfo/ros-users>
>>
>>
>>
>>
>>
>> ______________________________**_________________
>> ros-users mailing list
>> ros-users at code.ros.org
>> https://code.ros.org/mailman/**listinfo/ros-users<https://code.ros.org/mailman/listinfo/ros-users>
>>
>
>
> ______________________________**_________________
> ros-users mailing list
> ros-users at code.ros.org
> https://code.ros.org/mailman/**listinfo/ros-users<https://code.ros.org/mailman/listinfo/ros-users>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ros.org/pipermail/ros-users/attachments/20120702/efc48e7a/attachment-0004.html>