You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I noticed that none of the major rmw implementations implements rmw_get_serialized_message_size. I wonder if that is the case because it works on the rosidl_runtime_c__Sequence__bound struct and not an actual instance of a message. I could not really grasp how the rosidl_runtime_c__Sequence__bound should be used and to what structure * data should be pointing to.
Implementation considerations
Maybe it would be helpful to add a function interface that works on an actual instance of a message, that should be more straightforward to implement, e.g. for cyclonedds similar to this code.
The text was updated successfully, but these errors were encountered:
Hm, the fact that none of them implement it means the API should probably be deprecated and removed. I think this was part of an early attempt to be able to send preallocated messages, but never got implemented. @mjcarroll can you comment here?
I think that is the origin of it. I think that the loaned message APIs came after and superceded this concept. I think we can safely deprecate and remove the existing APIS.
Second, I think that the idea of working on a message instance may work as well. The Sequence__bound structure was intended to provide bounds for unbounded sequence fields so that allocation size could be correctly computed.
Feature request
Feature description
I noticed that none of the major rmw implementations implements
rmw_get_serialized_message_size
. I wonder if that is the case because it works on therosidl_runtime_c__Sequence__bound
struct and not an actual instance of a message. I could not really grasp how therosidl_runtime_c__Sequence__bound
should be used and to what structure* data
should be pointing to.Implementation considerations
Maybe it would be helpful to add a function interface that works on an actual instance of a message, that should be more straightforward to implement, e.g. for cyclonedds similar to this code.
The text was updated successfully, but these errors were encountered: