Disclaimer: Do not try this at home!

There is an ongoing DRC Ebola campaign where mappers are asked to update residential areas. All newly published projects follow the same pattern. There are a lot of small tasks to be mapped. The tasks are so small that no new residential areas need to be mapped in them. Only the existing mapping needs to be updated (move or delete buildings based on the imagery required by the project). Often the task area is completely empty.

Since the deployment of the v5.4.1 release and the panel rearrangement, completing a task takes more clicks than in the previous version. Even if you’re an experienced mapper who read the instructions before you started, the Instructions tab still opens first, so you have to switch to the Completion tab and scroll past the additional instructions to submit the task. Considering the small tasks in the campaign, the time it takes to submit a single task is more than half of the total time spent on mapping.

This is where laziness kicks in. And so does the Tasking Manager API.

The goal? Find patterns and use whatever means to speed up the process.

Prerequisites that must be met:

  1. No other mappers are active on the project. This matters because you work within the area defined by the project’s task grid without locking the tasks.
  2. You upload regularly to prevent conflicts (a mapper who locks a task in the Tasking Manager always takes priority over any changes you made there without locking it).

Validators may be suspicious when so many tasks are marked as mapped within seconds of each other. This diary documents the process and explains why these tasks met all the requirements to be marked as mapped.

The technical workflow in JOSM:

  1. Download the project’s task grid (using the TM API). Each task boundary has its ID and status.
  2. Using the search function, find the right number of small unmapped tasks with the query taskStatus=READY waylength:-300 (only the unmapped tasks with a perimeter under 300 meters).
  3. In the Tags/Memberships window, right-click the <X different> value of the taskId key and copy the list of values into a text editor for later.
  4. Use Download along to download all OSM data for the selected tasks as a new layer.
  5. Now add the task boundaries to the TODO list.
  6. Inspect the tasks one by one and mark each one as done in the TODO list.
  7. Upload the data to the OSM server (don’t forget to fill in the project’s changeset comment).

The Python script (outside JOSM):

  1. Use the project ID and the saved taskId list as input.
  2. A small script iterates over the taskId list and calls the TM API to lock each task and mark it as mapped (two API calls per task).
  3. If the script fails to lock a task, it skips it and moves on to the next one.

That completes the first iteration, and you can start over. The project’s task grid needs to be downloaded again to get the updated task statuses. Change the search query from waylength:-300 to waylength:-620. The rest of the steps remain the same. Don’t forget to update the taskId list for the script. The third iteration uses waylength:-950.

The result

After three iterations, 70% of the project was mapped and I saved roughly 5,000 mouse clicks and a great deal of time. At the time of writing, I had already used this workflow on five projects and mapped over 4,500 tasks. I didn’t encounter any merge conflicts.

Notes

I found that, for me, the sweet spot is around 200–400 tasks per batch. You can tweak the waylength value in the search query (all tasks smaller than…) to achieve the right number of tasks in the selection. The biggest slowdown was the OpenAerialMap imagery used in the Ebola campaign because its tiles load quite slowly.

When I only deleted features in a task, I just moved to the next task and uploaded the data every 100 tasks. If another mapper who locked the task had already deleted the same object, JOSM showed a warning that it no longer existed on the server, and the upload continued. If I added any features to a task, I uploaded the data immediately so my changes were already on the server in case someone else locked the task.

Conclusion

On every project, I stopped once all tasks matching waylength:-950 were done. Larger tasks take longer to update, so bypassing the Tasking Manager brings little benefit, and I suggest returning to the regular workflow with task locking.

Discussion

Comment from bluestarfish on 22 October 2026 at 16:37

This is legendary!:)

Leave a comment

Log in to leave a comment