backlog.md

Backlog

See also the ideas backlog.

Key:

• 🙏 Target for MVP
• 💩 Technical debt (⚠️ = important)

Features: core

✅ All done.

Features: nice to have

• Map sheet names
• Child hills relation (with pagination)
• Pagination enhancements
  • Allow skipping pages by number, rather than traversing cursors
  • Return total number of records
• "In" multi-valued filtering operator, to complement the single-valued equals and includes

Documentation

• Default query in the playground (needs graphql-playground/866)

Security

Keep it unauthenticated; API keys would be a barrier to usage. Revisit if abuse is a problem.

• Sanitise returned error messages (except for helpful GraphQL ones e.g. syntax error in query)

Performance

• Indices on filterable/sortable database fields
• Set ETags based on query & variables (not the response), to allow returning not-modified without running the database query
• 💩 Refactor database columns of type SET into multiple fields or separate table (FIND_IN_SET requires a table scan)

Operational

Production

• Domain

Monitoring

• ✅ AWS budget alarms

Technical

Move from DynamoDB to Aurora

• 💩 Replace type definitions in population script with type inference (as done on the query side)

Improvements

• Consider best practice for nullability
• Fully configure rules for ESLint Jest plugin
• Node.js 12 (LTS) is available on AWS Lambda as of November 2019. Switch to run on it, then convert to ES6 imports/exports (update lint rules), use promisified fs module, anything else new.
• Structured logging into CloudWatch, there must be a way
• Remove need to manually copy ARNs into database deployment scripts (needs serverless-pseudo-parameters#25) and Serverless offline environment variables (needs serverless-offline#388).

Further future

• Do without Express (?); it's only used to allow express-graphql to be used. Needs express-graphql/559, or manual implementation of basic requirements for GraphQL over HTTP.