bwi-update --sync failure

[jack-oquin]

This was run on a former Lucid workstation that was recently upgraded to Precise. There was not much installed before running bwi-update with the --system option.

$ bwi-update --sync
[Installing gazebo models in /home/joq/ros/gazebo]
requesting all changes
adding changesets
adding manifests
adding file changes
added 96 changesets with 664 changes to 393 files (+3 heads)
updating to branch default
345 files updated, 0 files merged, 0 files removed, 0 files unresolved
 - Done
[Initializing workspace of type: catkin in /home/joq/ros/catkin_ws]
/bin/bash: wstool: command not found
[Removing workspace from /home/joq/ros/catkin_ws]
 - Workspace does not exist in /home/joq/ros/catkin_ws
[Initializing workspace of type: rosbuild in /home/joq/ros/rosbuild_ws]
ERROR in config: Is not a local file, nor a valid URL [/home/joq/ros/catkin_ws/devel] : unknown url type: /home/joq/ros/catkin_ws/devel
[Removing workspace from /home/joq/ros/rosbuild_ws]
 - Workspace does not exist in /home/joq/ros/rosbuild_ws
[Merging https://raw.github.com/utexas-bwi/segbot/master/rosinstall/segbot.rosinstall in /home/joq/ros/rosbuild_ws]
Traceback (most recent call last):
  File "/usr/bin/bwi-update", line 612, in <module>
    error_code = synchronizeUserSpace()
  File "/usr/bin/bwi-update", line 516, in synchronizeUserSpace
    rosinstallROSPackage(rosinstall_target,build_type,ws[build_type])
  File "/usr/bin/bwi-update", line 334, in rosinstallROSPackage
    success = callWsCommand(["merge","--confirm-all","--target-workspace="+ws,rosinstall_file],type,ws)
  File "/usr/bin/bwi-update", line 139, in callWsCommand
    return call(["/bin/bash","-c","unset ROS_WORKSPACE && " + command + " " + " ".join(args)],cwd=ws,verbose=verbose,verbose_error=verbose_error)
  File "/usr/bin/bwi-update", line 124, in call
    p = subprocess.Popen(command, stdout=subprocess.PIPE, cwd=cwd)
  File "/usr/lib/python2.7/subprocess.py", line 679, in __init__
    errread, errwrite)
  File "/usr/lib/python2.7/subprocess.py", line 1249, in _execute_child
    raise child_exception
OSError: [Errno 2] No such file or directory: '/home/joq/ros/rosbuild_ws'

[jack-oquin]

Manually installing python-wstool works around the crash.

[piyushk]

Yeah. It was an exceptionally nasty way for crashing for a missing wstool.
I'll fix it in the next release. I must have missed python-wstool in my
config file.

Piyush

On Mon, Jan 14, 2013 at 9:26 PM, jack-oquin notifications@github.comwrote:

Manually installing python-wstool works around the crash.

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/4#issuecomment-12224520.

[jack-oquin]

Things like that are easy to overlook. That's why installing on a relatively clean system is a good test case.

[piyushk]

Absolutely true. I'll setup a vm and test it out.

On Mon, Jan 14, 2013 at 10:03 PM, jack-oquin notifications@github.comwrote:

Things like that are easy to overlook. That's why installing on a
relatively clean system is a good test case.

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/4#issuecomment-12226531.

[jack-oquin]

While updating my .bashrc (by hand), I noticed what looks like an error in ~/.bwi/workspace.bash:

source /home/joq/ros/rosbuild_ws/devel/setup.bash

I think that line should be:

source /home/joq/ros/rosbuild_ws/setup.bash

Unlike catkin workspaces, rosbuild workspaces have no devel subdirectory.

[piyushk]

I think I messed up. This looks like a genuine bug. Can you file it as a
separate issue? I'll get to it tomorrow.

Piyush

On Mon, Jan 14, 2013 at 10:10 PM, jack-oquin notifications@github.comwrote:

While updating my .bashrc (by hand), I noticed what looks like an error
in ~/.bwi/workspace.bash:

source /home/joq/ros/rosbuild_ws/devel/setup.bash

I think that line should be:

source /home/joq/ros/rosbuild_ws/setup.bash

Unlike catkin workspaces, rosbuild workspaces have no _devel_subdirectory.

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/4#issuecomment-12226875.

[jack-oquin]

Done, see: issue #5

[jack-oquin]

It took me two tries to compile:

$ rosmake gazebo_plugins segbot_apps segbot_gazebo

The first attempt failed, because I had not yet initialized rosdep:

$ sudo rosdep init
$ rosdep update

[piyushk]

Good catch. I forgot to call the functions responsible for the rosdep
update.

On Mon, Jan 14, 2013 at 11:11 PM, jack-oquin notifications@github.comwrote:

It took me two tries to compile:

$ rosmake gazebo_plugins segbot_apps segbot_gazebo

The first attempt failed, because I had not yet initialized rosdep:

$ sudo rosdep init
$ rosdep update

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/4#issuecomment-12229931.

[jack-oquin]

When I try to run the simulation, gazebo waits indefinitely:

waitForService: Service [/gazebo/set_physics_properties] has not been advertised, waiting

I ended up having to use the system monitor to kill the gzserver process. It would not stop on its own.

[piyushk]

I'll try to test this out tomorrow. I am not sure why this is happening.

Piyush

On Mon, Jan 14, 2013 at 11:26 PM, jack-oquin notifications@github.comwrote:

When I try to run the simulation, gazebo waits indefinitely:

waitForService: Service [/gazebo/set_physics_properties] has not been advertised, waiting

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/4#issuecomment-12230690.

[jack-oquin]

Me, neither. I updated the software on my Precise laptop, which had previously been set up and was working fine. It still works. That system defines a great many /gazebo/ services which are missing on the workstation I just set up.

I did notice the gzserver process not terminating properly on the laptop, too. I'll open a bug report on that specific issue (#6). Once it happens, gazebo stops launching properly.

[piyushk]

I'll try to debug this tomorrow. I suspect this is unrelated to bwi-update.
Thanks a lot for putting in time testing this tool. I am out for the night.

Piyush

On Tue, Jan 15, 2013 at 1:08 AM, jack-oquin notifications@github.comwrote:

Me, neither. I updated the software on my Precise laptop, which had
previously been set up and was working fine. It still works.

I did notice the gzserver process not terminating properly on the
laptop, too. I'll open a bug report on that specific issue. Once it
happens, gazebo stops launching properly.

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/4#issuecomment-12235643.

[jack-oquin]

I discovered I can reproduce this failure on my new desktop install just following the Starting Gazebo tutorial on ros.org:

$ source /opt/ros/groovy/setup.bash
$ roslaunch gazebo_worlds empty_world.launch

I don't think there is any BWI code involved in that test.

[piyushk]

I've pushed fixes to #5,#7 to my ppa. I'll be making one final cosmetic
change in a few hours - change the config file format to accept multiple
workspaces instead of hardcoding rosbuild/catkin/gazebo in the
configuration file.

By doing this I believe the configuration file format should be pretty
good. At the same time, I am not impressed by the way the tool is written.
I'll probably get back to re-writing the tool properly in a couple of
months for a potential wider release. More pressing things to work on right
now.

On Tue, Jan 15, 2013 at 2:00 AM, jack-oquin notifications@github.comwrote:

I discovered I can reproduce this failure on my new desktop install just
following the Starting Gazebo tutorialhttp://www.ros.org/wiki/simulator_gazebo/Tutorials/StartingGazeboon
ros.org:

$ source /opt/ros/groovy/setup.bash
$ roslaunch gazebo_worlds empty_world.launch

I don't think there is any BWI code involved in that test.

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/4#issuecomment-12238081.

[jack-oquin]

It seems to work OK.

I tend to blame my remaining problems on having done a dist-upgrade from Lucid to Precise, rather than a clean, new install. I am closing this issue, now.