I think I had the same problem running rosinstall recently. It was actually really hard to kill the zombie processes. But I didn't look into it more than that, and once ros was installed, it wasn't a problem again. Good to see there's a fix! On Mon, Nov 14, 2011 at 8:51 AM, Benjamin Charrow wrote: > Hi all. > > I've been experiencing a slight problem with rosdep and macports. > Basically, after I run rosdep, a macports process consumes 100% of my CPU > and commands I enter in the terminal aren't displayed (stdin is stolen). I > have macports 2.0.3 and am using the ros-1.6.6 tagged version of ros. Has > anyone else had this problem? > > Looking at the source, the issue seems to be a check for whether or not > macports exists in the constructor of rosdep.installers.MacportsInstaller. > There's a popen command which opens up a macports session, but the process > doesn't look like it's being killed. This would explain why I/O is being > stolen, though I'm not sure why my CPU usage spikes. My quick fix was to > just change the command to be "port help" (see diff bellow). > > Should I file a bug report? > > Cheers, > Ben > > SVN URL: > https://code.ros.org/svn/ros/stacks/ros/tags/ros-1.6.6/tools/rosdep/src/rosdep > > bcharrow@amateur:/opt/ros/electric/ros/tools/rosdep/src/rosdep > $ svn diff > Index: installers.py > =================================================================== > --- installers.py (revision 15358) > +++ installers.py (working copy) > @@ -424,7 +424,7 @@ > packages = packages.split() > > try: > - pop = subprocess.Popen(['port'], stdout=subprocess.PIPE, > stderr=subprocess.PIPE) > + pop = subprocess.Popen(['port', 'help'], > stdout=subprocess.PIPE, stderr=subprocess.PIPE) > except OSError: > import traceback; traceback.print_exc() > print("There was an error running macports, make sure it is > installed properly.", file=sys.stderr) > _______________________________________________ > ros-users mailing list > ros-users@code.ros.org > https://code.ros.org/mailman/listinfo/ros-users >