Build more than one region and view them together on one map.
bash quickstart.sh builds one region per run, and each run’s map shows
only that region. Choose a method:
| Regions | Method |
|---|---|
| A few small ones, for checking | On your computer |
| A town or district | On the shared VM |
| Many districts | On a worker fleet (experimental) |
tools/find_area.py; see
How to build a region.find_area.py --details).Build each region:
bash quickstart.sh queenstown-lakes-district/arrowtown/arrowtown --no-open-browser
bash quickstart.sh queenstown-lakes-district/queenstown/sunshine-bay --no-open-browser
Check each run’s report.md. Stop each preview server that the pipeline
starts. The local quickstart shows how.
Combine them into one map:
.venv/bin/python src/combine_regions.py --skip-markup-lines --regions \
queenstown-lakes-district/arrowtown/arrowtown \
queenstown-lakes-district/queenstown/sunshine-bay
This replaces the map data in data/ with these regions. If you leave out
--regions, every region under data/out/ is included.
Serve and open it:
.venv/bin/python tools/pipeline_serve.py --root . --port 8000
Open http://localhost:8000/rooftop-solar.html. Press Ctrl+C to stop.
This uses billable VM time. Unlike the cloud quickstart, remote_build.sh
shows no VM cost estimate. Estimate the cost with the
Google Cloud pricing calculator.
The estimate is not a spending cap. Agree it with whoever pays before you
run the build.
Push your code to main, or pass --ref <branch>.
Build, gate and publish a developer preview:
tools/remote_build.sh --fetch --preview-user <github-alias> --regions \
"queenstown-lakes-district/arrowtown/arrowtown queenstown-lakes-district/queenstown/arrow-junction"
Answer the paid-referee question. Enter declines it. The VM starts if it is stopped, and stops at the end if this run started it.
If the run is cut off, run the same command again. It continues from where it stopped.
Open
https://storage.googleapis.com/rewiring-solar-previews/devel/<github-alias>/rooftop-solar.html
and read any GATE WARNING list.
The combined map includes every region that has been emitted on the VM, not
only the regions in --regions. A production release uses --publish and
the deploy gate, not --preview-user.
You can also start this workflow from GitHub: go to Actions, then
Build and publish the map, and enter space-separated regions.
Experimental: multi-worker throughput and recovery are not yet proven. Each worker creates a billable spot VM. Read Many machines and agree the cost before you run:
.venv/bin/python tools/enqueue_regions.py --config --dry-run # what would be queued
.venv/bin/python tools/enqueue_regions.py --config # queue config.REGIONS
tools/fleet.sh up 2 # start 2 workers
.venv/bin/python tools/status.py # progress
tools/fleet.sh down # delete every worker
Workers publish each region’s outputs. They do not combine the regions or release a map.