[jsk_topic_tools] Function to warn with no remappings

[wkentaro]

No description provided.

[k-okada]

this is general comment on topic tools

• using python on this type of topic tools may not a good idea, especially
  for simulated environment; they have significant performance issues
  discussed at
  http://answers.ros.org/question/29425/rospy-horrible-performance-with-sim_time/
  (btw, kwc:ken is the original rospy coder, JSK mentor during pr2 program,
  and moved to G 1-2 year before they acquired many robotics company.), so
  if we use this kind of highspeed message relay tools, we'd better to choose
  c++ client (or roseus which call c++ function internally), or dig into the
  python implementation and find the solution. of course, this is the lowest
  priority task.
  Cf.
  hrpsys_ros_bridgeを立ち上げるとCPUパワーが１００％になる start-jsk/rtmros_common#692 (comment)

◉ Kei Okada

On Wed, Nov 11, 2015 at 3:29 AM, Kentaro Wada [email protected]
wrote:

---

You can view, comment on, or merge this pull request online at:

#1224
Commit Summary

• [jsk_topic_tools] Function to warn with no remappings

File Changes

• M jsk_topic_tools/src/jsk_topic_tools/init.py
  https://github.com/jsk-ros-pkg/jsk_common/pull/1224/files#diff-0 (1)
• A jsk_topic_tools/src/jsk_topic_tools/validation_util.py
  https://github.com/jsk-ros-pkg/jsk_common/pull/1224/files#diff-1
  (24)

Patch Links:

• https://github.com/jsk-ros-pkg/jsk_common/pull/1224.patch
• https://github.com/jsk-ros-pkg/jsk_common/pull/1224.diff

—
Reply to this email directly or view it on GitHub
#1224.

[wkentaro]

• using python on this type of topic tools may not a good idea, especially
  for simulated environment; they have significant performance issues
  discussed at

Where is bottleneck? here?

if we use this kind of highspeed message relay tools

I think this is not relay tool precisely, do you say 'relay' as example?

[wkentaro]

• using python on this type of topic tools may not a good idea, especially for simulated environment; they have significant performance issues discussed at

and why do you specify the situation use_sim_tim:=true ?
because each node has to listen /clock topic?

[wkentaro]

% rosparam get /use_sim_time
ERROR: Parameter [/use_sim_time] is not set

% echo "$(rosnode list | wc -l) , $(rostopic list | wc -l)"
94 , 495

% ./warn_no_remap_performance.py
rospy.get_name
elapsed: 2e-06 [s]
rospy.names.resolve_name
elapsed: 1.1e-05 [s]
['/spam/input', '/spam/input/a', '/spam/output', '/spam/output/a']
rospy.names.get_resolved_mappings
elapsed: 1e-06 [s]
{}

warn_no_remap.py

[wkentaro]

if we use this kind of highspeed message relay tools, we'd better to choose
c++ client (or roseus which call c++ function internally), or dig into the
python implementation and find the solution. of course, this is the lowest
priority task.

This is a tool when writing node of python, so we cannot use c++ or roseus as the substitution.
So if the performance is too critical, I need change rospy implementation. (maybe call c++ function inside it if currently not)

C++ version for this is #1229

[k-okada]

do you say 'relay' as example?

yes, and he said listing topic running at 1000hz may consume CPU power,
which may not become a problem on C++ implementation. so I believe you want
to write everything in python, but you may need to fix this problem at
some point, not now.

◉ Kei Okada

On Wed, Nov 11, 2015 at 10:45 PM, Kentaro Wada [email protected]
wrote:

• using python on this type of topic tools may not a good idea,
  especially for simulated environment; they have significant performance
  issues discussed at

and why do you specify the situation use_sim_tim:=true ?
because each node has to listen /clock topic?

—
Reply to this email directly or view it on GitHub
#1224 (comment)
.