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
We were attempting to add visualization_msgs to be generated for C#, but noticed that the generated C# code is expecting Header to be a message within visualization_msgs and not the std_msgs/Header. When visual studio tries to compile the generated C# code, it will fail because no visualization_msgs/Header exsits.
We have found out that this appears to depend on the alphabetical order of the package that the messages are in. If we rename visualization_msgs to avisualization_msgs (or have it start with any letter before s) everything will generate and compile correctly, and if it comes after std_msgs it will experience the error described above.
It looks like we can get everything to build correctly if we modify the messages themselves to have std_msgs/Header instead of Header in the .msg definition. But this is not ideal since it requires us to maintain a different & parallel message definition, and I believe will also cause the md5 of the messages to not match.
The text was updated successfully, but these errors were encountered:
If you're curious, non-literal fields' types' md5 sums are inserted in place of their type names during md5 calculation, allowing explicit or implicit package determination to work).
I hadn't noticed the alphabetical component, but that would definitely explain why sensor_msgs/Images can simply use "Header" with no package name, but trajectory_msgs cannot! Thanks!
We were attempting to add
visualization_msgs
to be generated for C#, but noticed that the generated C# code is expectingHeader
to be a message withinvisualization_msgs
and not thestd_msgs/Header
. When visual studio tries to compile the generated C# code, it will fail because novisualization_msgs/Header
exsits.We have found out that this appears to depend on the alphabetical order of the package that the messages are in. If we rename
visualization_msgs
toavisualization_msgs
(or have it start with any letter befores
) everything will generate and compile correctly, and if it comes afterstd_msgs
it will experience the error described above.It looks like we can get everything to build correctly if we modify the messages themselves to have
std_msgs/Header
instead ofHeader
in the .msg definition. But this is not ideal since it requires us to maintain a different & parallel message definition, and I believe will also cause themd5
of the messages to not match.The text was updated successfully, but these errors were encountered: