Coastal

Coastal builds a cycling dashboard for the coastal ride project. It consumes ride activity and stream data from the cycling_platform_silver layer, combines it with hand-maintained coastal segment metadata, and renders the public dashboard to root index.html.

The generated root index.html file is the GitHub Pages publishing artifact. Do not commit regenerated dashboard output while validation is failing.

Current Status

The project is currently blocked on an upstream silver streams issue: cycling_platform_silver.activity_streams is expected to contain stream rows for the coastal activity IDs, but the most recent smoke check found it empty.

Before publishing:

Rscript scripts/check_inputs.R

Entry Points

• index.Rmd: renders the flexdashboard and writes root index.html.
• ride_explorer.R: inspects one activity stream to determine coastal crop windows.
• scripts/check_inputs.R: validates local metadata against the silver tables.
• coastal_visualisation.R: creates static rider visualisations.
• R/rider_visualisation.R: builds and exports testable rider poster plots.
• reverse_geocoder.R: supports postcode/location enrichment.

Project Layout

• R/load.R: loads all project modules in dependency order.
• R/database.R: reads database settings and creates explicit connections.
• R/dependencies.R: checks and attaches runtime packages at entry points.
• R/metadata.R: loads metadata and shared constants.
• R/validation.R: validates ride and ferry metadata.
• R/silver_streams.R: owns silver stream column selection and stream loading.
• R/coastal_data.R: builds the core coastal dataset and ride summaries.
• R/geo_images.R: handles position extremities and image metadata.
• R/maps.R: renders maps and rider trace PNGs.
• R/plots.R: renders dashboard plots.
• R/valueboxes.R: renders coordinate value boxes.
• data/coastal_activities.R: hand-maintained coastal ride segment metadata.
• data/ferries.R: hand-maintained ferry marker metadata.

Use this in project entry points:

source("R/load.R")

R/load.R only sources project modules. Entry points explicitly load their runtime packages and create a database connection:

source("R/load.R")
load_coastal_packages()
con <- connect_coastal_database()

Local Configuration

connect_coastal_database() reads database settings from environment variables. Keep secrets in .Renviron, which is ignored by git. Start from .Renviron.example.

Required variables:

CYCLING_PLATFORM_SILVER_SCHEMA=cycling_platform_silver
MARIADB_HOST=cycling-prod.local
MARIADB_PORT=3306
MARIADB_USER=your_user
MARIADB_PASSWORD=your_password

CYCLING_PLATFORM_SILVER_SCHEMA selects the schema used by every silver table query. The connection itself does not require a default database name.

Add A Ride

1. Add the ride segment to data/coastal_activities.R in geographic order.
2. Use ride_explorer.R to inspect the silver activity stream and choose ride_start_time / ride_end_time.
3. Add ferry markers to data/ferries.R if relevant.
4. Run Rscript scripts/check_inputs.R.
5. Run Rscript scripts/render_dashboard.R.
6. Review and commit any source changes separately.

Publishing

Run scripts/render_and_publish.sh to validate inputs, render index.Rmd to root index.html, then commit and push the generated dashboard. Each step must succeed before the next begins.

For a render without publishing, run Rscript scripts/render_dashboard.R.

Do not publish while scripts/check_inputs.R fails. In the current state, publishing should wait until the silver activity_streams table is populated.

Rider Visualisations

Run Rscript coastal_visualisation.R to create one coastal-only poster per rider. Outputs are written to outputs/visualisations/; interactive route previewing is disabled when run with Rscript.

Static Checks

Parse the project R files without querying the database:

Rscript -e "files <- c(list.files('R', full.names = TRUE), 'scripts/check_inputs.R', 'ride_explorer.R', 'coastal_visualisation.R', 'reverse_geocoder.R'); invisible(lapply(files, parse))"

Check the R Markdown chunks:

Rscript -e "out <- knitr::purl('index.Rmd', output = tempfile(fileext = '.R'), quiet = TRUE); invisible(parse(file = out))"

Run the automated tests:

Rscript tests/testthat.R

More Documentation

• docs/data-contract.md
• docs/ride-maintenance.md
• docs/development.md
• docs/publishing.md