# Data on most popular ROS packages

**URL:** <https://discourse.openrobotics.org/t/data-on-most-popular-ros-packages/13591>\
**Category:** ROS General\
**Created:** [April 9, 2020, 1:30pm UTC](https://discourse.openrobotics.org/t/data-on-most-popular-ros-packages/13591 "2020-04-09T13:30:45Z")\
**Posts on this page:** 7\
**Page:** 1

<div class="post-metadata">

**Author:** ![roycey](https://avatars.discourse-cdn.com/v4/letter/r/9fc29f/32.png) [@roycey](https://discourse.openrobotics.org/u/roycey)\
**Post date:** [April 9, 2020, 1:30pm UTC](https://discourse.openrobotics.org/t/data-on-most-popular-ros-packages/13591/1 "2020-04-09T13:30:45Z")

</div>

In the [Metrics](http://metrics.ros.org/) report released late last year , the top 40 most downloaded packages were mentioned, however, they are the default packages available with ROS distros. What is the right source for me to look for the most popular and commonly used ROS packages?

During my search, I came across a thread that stated that MoveIt is the 3rd most popular package. Is this from a report? If yes, please do share.

Secondly, I was looking at the OS related metrics for ROS2 [here](https://metrics.ros.org/packages_ros2.html) How do I interpret the popularity of Windows over Linux? Is the numbers skewed because ROS2 on Linux is installed via apt?

---

<div class="post-metadata">

**Author:** ![gavanderhoorn](https://sea2.discourse-cdn.com/flex022/user_avatar/discourse.openrobotics.org/gavanderhoorn/32/969_2.png) [@gavanderhoorn](https://discourse.openrobotics.org/u/gavanderhoorn)\
**Post date:** [April 9, 2020, 2:21pm UTC](https://discourse.openrobotics.org/t/data-on-most-popular-ros-packages/13591/2 "2020-04-09T14:21:09Z")

</div>

I believe some of this was discussed in the later comments on [ROS Metrics - ROS General - Open Robotics Discourse](http://discourse.ros.org/t/ros-metrics/10861).

Some of the plots are somewhat skewed because of the way the source data captures events.

@DLu and @tfoote would be able to explain this better.

---

<div class="post-metadata">

**Author:** ![DLu](https://sea2.discourse-cdn.com/flex022/user_avatar/discourse.openrobotics.org/dlu/32/19014_2.png) [@DLu](https://discourse.openrobotics.org/u/DLu)\
**Post date:** [April 9, 2020, 2:46pm UTC](https://discourse.openrobotics.org/t/data-on-most-popular-ros-packages/13591/3 "2020-04-09T14:46:29Z")

</div>

There’s no great way to determine what’s most commonly used. There’s only some indirect proxies.

[Most Downloaded Packages](https://metrics.ros.org/packages_top.html) is one way to measure, but that’s based on downloading package binaries from [packages.ros.org](http://packages.ros.org), which not everyone does, and doesn’t count if you downloaded from source. The website only lists the top ten for aesthetic reasons. I could post the current full list if desired.

[The Repos Table](https://metrics.ros.org/repos_table.html) sorts based on Github metrics (forks/stars/subs)

The ROS2 data was discussed on [Discourse previously](http://discourse.ros.org/t/2019-ros-metrics-report/10938/4). I cannot speak to the difference between these numbers and the ones from [packages.ros.org](http://packages.ros.org). If there are different ways of visualizing this data could be useful, please let me know by [opening an issue](https://github.com/DLu/ros_metrics/issues/new)

---

<div class="post-metadata">

**Author:** ![smac](https://sea2.discourse-cdn.com/flex022/user_avatar/discourse.openrobotics.org/smac/32/6763_2.png) [@smac](https://discourse.openrobotics.org/u/smac)\
**Post date:** [April 9, 2020, 6:37pm UTC](https://discourse.openrobotics.org/t/data-on-most-popular-ros-packages/13591/4 "2020-04-09T18:37:55Z")

</div>

> [@DLu](#):
>
> and doesn’t count if you downloaded from source

Also won’t count docker or snap containers distributed for production.

> [@DLu](#):
>
> I could post the current full list if desired.

I’d also be interested in having an annual complete list or adding to your metrics repo table. Though on the bottom of the page it says: “the most downloaded packages introduced in each year”. I’m more interested in the top packages of the year regardless of their introduction, though having a top-10 of new things is a nice metric too!

---

<div class="post-metadata">

**Author:** ![DLu](https://sea2.discourse-cdn.com/flex022/user_avatar/discourse.openrobotics.org/dlu/32/19014_2.png) [@DLu](https://discourse.openrobotics.org/u/DLu)\
**Post date:** [April 10, 2020, 2:00am UTC](https://discourse.openrobotics.org/t/data-on-most-popular-ros-packages/13591/5 "2020-04-10T02:00:10Z")

</div>

[Full List of Top Packages](https://gist.github.com/DLu/0ce75ef609f78d0bf171b9dc6bcf0477)

5308 packages in total. Fun fact: I’m responsible for one of the least downloaded packages ever: [`joint-state-publisher-js`](https://wiki.ros.org/joint_state_publisher_js/Tutorials/Using%20JointStatePublisher%20in%20the%20Browser)

For the latter, I opened [a ticket](https://github.com/DLu/ros_metrics/issues/23) to make a separate page for that. I had a vague memory of the top packages not changing too much, but that might have been from before I got the expanded logs. There’s definitely some variability. I’ve been looking at [zingchart](https://www.zingchart.com/docs/chart-types/rankflow) for displaying a couple of unique types of data, like these rankings and the [country rankings](https://metrics.ros.org/packages_country.html)

```auto
---- - --------------------- -------
2014 0 std-msgs 52229
2014 1 message-runtime 51361
2014 2 rosconsole 51196
2014 3 xmlrpcpp 51119
2014 4 rosgraph-msgs 50987
2014 5 roscpp 50574
2014 6 geometry-msgs 50393
2014 7 rospy 50381
2014 8 catkin 50367
2014 9 sensor-msgs 50307
2015 0 roscpp 87901
2015 1 catkin 87470
2015 2 tf 85820
2015 3 rosbag 85678
2015 4 rosbuild 84776
2015 5 rosconsole 84572
2015 6 std-msgs 84555
2015 7 rosgraph-msgs 84479
2015 8 message-generation 84446
2015 9 cpp-common 84376
2016 0 urdf-tutorial 305533
2016 1 roscpp 289902
2016 2 message-runtime 289460
2016 3 message-generation 289024
2016 4 genpy 288653
2016 5 std-msgs 287318
2016 6 rospack 286020
2016 7 xmlrpcpp 285658
2016 8 rosgraph-msgs 285393
2016 9 roslib 284745
2017 0 rviz 505440
2017 1 tf2-ros 498131
2017 2 tf 494169
2017 3 tf2-py 475799
2017 4 gazebo-plugins 456251
2017 5 laser-geometry 453742
2017 6 actionlib 448684
2017 7 tf2-msgs 447002
2017 8 interactive-markers 446854
2017 9 tf2 445335
2018 0 tf2-ros 819624
2018 1 tf 807983
2018 2 tf2-py 770879
2018 3 tf2-msgs 770595
2018 4 tf2 768915
2018 5 pcl-ros 732875
2018 6 cv-bridge 716967
2018 7 actionlib 711472
2018 8 laser-geometry 710910
2018 9 rviz 710391
2019 0 tf2-ros 1130085
2019 1 actionlib 1123773
2019 2 resource-retriever 1122357
2019 3 rviz 1120802
2019 4 pluginlib 1114987
2019 5 robot-state-publisher 1101370
2019 6 sensor-msgs 1097838
2019 7 tf 1082884
2019 8 message-filters 1079182
2019 9 roscpp 1073007
2020 0 rviz 518233
2020 1 ros-core 515827
2020 2 ros-base 514639
2020 3 ros-comm 488302
2020 4 tf2-ros 485901
2020 5 roslisp 482499
2020 6 tf 475840
2020 7 resource-retriever 468266
2020 8 sensor-msgs 467686
2020 9 desktop 466699
---- - --------------------- -------

```

---

<div class="post-metadata">

**Author:** ![smac](https://sea2.discourse-cdn.com/flex022/user_avatar/discourse.openrobotics.org/smac/32/6763_2.png) [@smac](https://discourse.openrobotics.org/u/smac)\
**Post date:** [April 10, 2020, 2:26am UTC](https://discourse.openrobotics.org/t/data-on-most-popular-ros-packages/13591/6 "2020-04-10T02:26:23Z")

</div>

This data is surprisingly interesting. @Katherine_Scott maybe something to play with your data science skills.

I’m surprised how high up some of the ROS2 stuff is given ROS1 has a 8 year head start.

This could make for some nice superlatives. Of the packages I made alone, my lowest is [ros2 ouster](https://gist.github.com/DLu/0ce75ef609f78d0bf171b9dc6bcf0477#file-top_report-txt-L5117) with 26 binary installs. Which is surprising to me knowing some folks that use it its that low. But its in the lowest 3.6% of installs (but hopefully growing). My highest is [STVL](https://gist.github.com/DLu/0ce75ef609f78d0bf171b9dc6bcf0477#file-top_report-txt-L1068) in the top 20% with nearly 16k.

This has some really interesting insights in it - like I’m seeing grid-maps scattered pretty high all over the place. This was an area I was interested in and migrating some elements of it to Navigation2, but it seems I’m late to the game, everyone else is already doing it!

---

<div class="post-metadata">

**Author:** ![davetcoleman](https://sea2.discourse-cdn.com/flex022/user_avatar/discourse.openrobotics.org/davetcoleman/32/1835_2.png) [@davetcoleman](https://discourse.openrobotics.org/u/davetcoleman)\
**Post date:** [April 15, 2020, 2:29am UTC](https://discourse.openrobotics.org/t/data-on-most-popular-ros-packages/13591/7 "2020-04-15T02:29:46Z")

</div>

Regarding MoveIt, in ROS’s 2014 survey the following question was posed:

> **Which ROS packages are most important to you?**

And MoveIt was ranked #3

> **[ROS user survey: the results are in - ROS robotics news](https://www.ros.org/news/2014/04/ros-user-survey-the-results-are-in.html)**

I have not seen a similar question asked since then to know if that is still the case.
