ROS Resources: Documentation | Support | Discussion Forum | Index | Service Status | ros @ Robotics Stack Exchange
Ask Your Question

catkin_prepare_release pushes twice w/o incrementing tag?

asked 2013-04-06 15:31:34 -0600

130s gravatar image

updated 2013-04-09 10:03:04 -0600

I ran catkin_prepare_release once for the metapackage I maintain, then in its commit history, you can see commit message 0.2.12 twice on the same timestamp. Looking into the changes in each commit, the version defined in package.xml is incremented so that the latest is 0.2.13. However the latest tag in the github repository still remains as 0.2.12.

To summarize what I did not intend:

  • Committed twice by running catkin_prepare_release only once
  • If committing twice is necessary, tag is not yet updated in the 2nd commit

Is this intended behavior? If so, what steps are the next? I haven't notified buildfarm since this looks not good to me.

I admit that I was working in public wifi which might not be recommended for critical tasks like this, but I don't think the connection wasn't too bad.


ros-groovy-catkin 0.5.65-3quantal-20130325-0722-+0000

Update) command suggested by catkin_prepare_release:

../../../../../../../../../opt/ros/groovy/bin/catkin_package_version --bump patch . && git commit -m "0.2.12" rqt_runtime_monitor/package.xml rqt_robot_steering/package.xml rqt_tf_tree/package.xml rqt_robot_dashboard/package.xml rqt_pose_view/package.xml rqt_nav_view/package.xml rqt_robot_plugins/package.xml rqt_rviz/package.xml rqt_robot_monitor/package.xml && git push && git tag 0.2.12 && git push --tags
edit retag flag offensive close merge delete


FYI, next time tag with catkin or something common otherwise I don't get notified on this question.

William gravatar image William  ( 2013-04-09 09:56:14 -0600 )edit

I would have to see the command that catkin_prepare_release gave you to run, it shouldn't commit twice.

William gravatar image William  ( 2013-04-09 09:57:47 -0600 )edit

The posted output will not commit twice, you must have run that output twice, I bet if you look at the history then the second 0.2.12 will have package.xml which read 0.2.13... And they do, so I would say you must have accidentally run that stuff twice.

William gravatar image William  ( 2013-04-09 10:05:26 -0600 )edit

Hm, history only returns a single line, but anyway sounds like it isn't a big deal. So I'll reset the last wrong commit (in this case 2nd 0.2.12). Do you want to re-post your last comment so that I'll mark it as an ans

130s gravatar image 130s  ( 2013-04-09 10:11:13 -0600 )edit

1 Answer

Sort by ยป oldest newest most voted

answered 2013-04-09 13:02:57 -0600

130s gravatar image

updated 2013-04-09 13:38:25 -0600

I found by myself that the version in package.xml gets incremented when catkin_package_version runs, even though authentication that subsequently comes after incrementation fails with the repository (github in my case). Since catkin_package_version repeats letting me try authentication, version can increment as many times as you fail + 1 success in the end.

I don't know if it's designed to do so or not but I opened a ticket.

edit flag offensive delete link more



Yes, it is unsafe in that way, it just uses && to string commands together, so the first command runs even if the second one fails.

William gravatar image William  ( 2013-04-09 13:04:28 -0600 )edit

Question Tools


Asked: 2013-04-06 15:31:34 -0600

Seen: 590 times

Last updated: Apr 09 '13