This is a web application that serves basic weather information given a requested Zip Code. To satisfy basic needs, it currently retrieves external weather data and caches results by zip code for display.

At this stage of functionality, no database is needed since the only persisted data is cached from api calls and renders. This decision keeps things simple initially, but could evolve to store fetched results in an application database potentially with enhanced data, longer history, etc. SQLite 3 is left as a dependency though could be removed if not planning to add functionality requiring a database.

Since the initial assumptions of the project were to cache results by zip code, address entry kept simple to bootstrap the process. Ultimately, the application could add searching based on Address, city, etc. to help users search a location if they don't know the zip code.

Since this first pass focuses on fetching results at the zip-code level only, we encounter the issue that zip codes are not unique across all geographic areas (mainly country). So at this stage, US zip code lookup is enforced. We make the assumption that cities that share a zip code in the US are geographically close such that results can be shared.

Currently the applicaiton displays current conditions, but is set up to make a second call for forecast information to render as well. For most free API plans this requires two calls though there are some paid plans that provide a single endpoint to get both sets of data.

All data is currently forced to imperial units, though with an extra API parameter and a bit more view logic this could become user selectable.

• With no database, there is no data recorded other than logging to track what locations are requested and how often
• Heavy reliance on caching both for the api results and views.

With the current simplicity of the user interface, there is a single Controller called WeatherController. It supports an index view taking user input (address information focusing on zip code at present).

The only model is LocationWeather, a client for the external weather API. It abstracts the details of making requests to the external weather API and handling types of responses at a high level. At present though, the response JSON is passed directly as a Hash for use in the views, so the abstractions currently leaks. As this application evolved, we would want to change this, most likely defining internal fields for data and mapping those within the client or a result object that can adapt the responses to the applicaiton standard (in order to decouple the particular API chosen from the rest of the application).

Due to caching and no database interaction, the application can likely handle reasonably significant traffic by providing a good CacheStore and sufficient pool of request handlers.

If there are a significant number of frequent requesters, adding expiring cache headers would improve caching further, but probably not for most indivdiual users. The next step here would be to store cached results in a proxy cache by setting headers to "public and cacheable" and fronting the applicaiton with a reverse proxy cache server such as nginx or even a CDN.

This project was set up to use the Tomorrow API. You will need to sign up and create a free account in order to get your own api key.

To configure the api key for the application, set the environment variablie TOMORROW_API_KEY so that it will be available to the Rails application.

To keep data retrieval and caching simple, only a zip code is needed. In the future if may be desireable to support more specific geolocation, which will have an impact on how much cache space is needed.

Run rails css:build

For production, make sure to run rails assets:precompile.

Weather data is cached for 30 minutes by zip code, along with the views that will not change until the data is refreshed.

To test caching in development mode, remember to enable caching with rails dev:cache

• Ruby version 3.3.x
• Rails 7.1.X

Tested with Ruby 3.3.0.

To minimize initial dependencies, the project relies on Rails default cache mechanisms which may not be suitable for certain deployment needs. To change, configure a different cache store in Rails.

Current Dependencies:

• Ruby 3.3.x
• Nodejs
• yarn
• sqlite3

bundle exec rails server

bundle exec rspec