GeoJupyter Collaboration Café

GeoJupyter Collaboration Café (2026-07-08)¶

Please add new agenda items under the New agenda items heading!

Your name / GitHub ID / affiliation / icebreaker

Matt / GitHub ID / affiliation / ?
Benny / GitHub ID / affiliation / ?
Maryam / GitHub ID / affiliation / ?
Name / GitHub ID / affiliation / ?

Agenda & notes¶

⚡ (5 minutes) Lightning intros¶

Tell us about you in 30 seconds or less!

🌐 (5 minutes) Lightning demo¶

What’s new? Show & tell. Post on Zulip to request a show & tell slot; by default, QuantStack will demo awesome JupyterGIS progress each meeting!

💡 (5 minutes) What will we work on?¶

What do you want to work on today?
- For ideas, check out the hackathon and good first issue labels on the JupyterGIS project!
Add your ideas to the “ideas” list below.
Add your favorite emoji or a + next to ideas you’re excited about.
- Press the colon (:) key on your keyboard or navigate to “Insert > Emoji” in the menu bar to open the emoji browser.

Action items¶

Matt: Make a sequence diagram of events explored below
Matt: Make a design fiction blog post including sequence diagram
Matt & Maryam: Review the above, then share it with the team (including Juliana, ?)!

Notes¶

Trail:
- Handling work across computers?
  - Persistent project ID in the trailfile. Use GitHub to clone the project to the other machine.
  - Analogy to lockfiles.
  - https://docs.sylabs.io/guides/3.5/user-guide/introduction.html#
- Collaboration with another person across different aspects of the same project?
  - Maryam: Haven’t thought about RTC yet -- ignore RTC for now
  - Maryam: Focus on single-user workflow for now.
- Emulating Windows on MacOS for certain tools?
  - We only care about tools in the Jupyter ecosystem
- If we pin this to Jupyter-only tools, do we actually have the tools in our ecosystem to enable people to stop using their non-Jupyter tools (like QGIS, JOSM, ?)
  - Maryam: Haven’t touched JGIS/QGIS/Arc for 4 years. Have been in Python instead! Jupyter Notebooks for viz/map export only.
- Service vs multiple processes writing to a single shared file (and possible collisions), multiplayer support?
  - Run in jupyter server?
- Very simple use case
  - Start from a project by Maryam’s students!
  - https://rabahbabaci.github.io/MTC-Transit_Access_Equity/
    - User is downloading data from a public network with a Jupyter data discovery tool
      - Mock JupyterDataConnect -- it outputs a file that was downloaded and a STAC file
      - Event: Query was performed.
      - Event: Data was acquired.
    - Imports into JGIS
      - Event?: User started a JGIS project
      - Event: Data was loaded in JGIS
      - Event: User zoomed in to <bbox> and lingered there for 45 seconds.
      - Event: User drew a polygon
      - Event: User buffered the polygon
    - Does cleaning steps in a Notebook
      - Event: User executed a Notebook cell that opens previously-acquired data. For now, we don’t care about linking this event to the data-acquisition event.
      - Event: User executed Notebook cell with streaming data. For now, code is the only provenance -- no static analysis, no relationship to other events -- e.g. user streamed some data? There’s an xarray.open_dataset(...) call that describes that.
    - Does map interactions
      - Event: For each major interaction
    - User uses another Notebook for exploration
      - TODO: Flesh this out later.
    - User uses another Notebook to finish analysis, producing outputs data files
      - Event: User runs Notebook cell that reads data. Only record what code was executed, we don’t care about what the code actually did.
      - Event: Intermediate data products are read. Trail is constantly watching the folder for read files (i.e. listening to file events). This event is not connected to the previous code cell.
      - Events: Notebook execution and file reads
      - Event: Artifact created in the project folder
        Trail is constantly watching the folder for new files (i.e. listening to file events). This event is not connected to any code cell events
    - Publication:
      - Print citations: When/how did the user acquire the data
      - Print methods: What code was executed
      - Write a MyST site
      - Event: myst build --html
      - If the user pushed to GitHub and GitHub builds the site, we don’t record anything. They can manually create an event if they want to.
  - https://storymaps.arcgis.com/stories/9ab0271a09d4475fbccc68f57e97d70e
  - https://anandashar01.github.io/bart-transit-equity-full/
- What does the actual protocol / message format look like?
  - Can we mock up an example trail file?
  - What expectations for tool implementation do have? Do we want those tools, like JGIS, to save provenance information internally in their own artifacts and write copies to Trail?
- Clearly defined MVP?