This is built using the brand new Zeit Now API, which makes it easy to write software that in turn deploys other software: https://zeit.co/blog/api-2
Has some really nice apis for group by, join and intutive way to perform 'where', 'order by' and other MySql clauses.
Here's his datasette: https://datasette-cmwqbwgnqe.now.sh/csv-data-5284def/nuforc_...
And here's an example SQL query showing the cities with the most UFO sightings: https://datasette-roborbllfq.now.sh/csv-data-b1429ef?sql=sel...
This is a good fit for any time you want to share some static data with the world - instead of posting a CSV file, you can share your data as a Datasette instance with a browseable web UI and a JSON API instead.
The three use-cases I'm personally most interested in for this are:
Data journalism: newspapers collect data about e.g. police shootings https://vice-police-shootings.now.sh/ or actions under the antiquities act https://fivethirtyeight.datasettes.com/fivethirtyeight-2628d... - I want to make it as easy as possible to publish the data beneath the stories.
Government open data: cities like San Francisco publish huge amounts if interesting data. I used San Francisco's 190,000 line CSV of trees in the city to build https://sf-tree-search.now.sh/
Cultural institutions: museums, libraries and art galleries have the collection and sharing of data as part of their mandates. I want to make it as easy as possible to build things like the London Science Museum's collection API: https://group.sciencemuseum.org.uk/about-us/collection/using...
No need for an app, app server, and the json can be cached with the rest of your content. Unless I’m missing something?
The site is built in Jekyll so it is 100% static. The dynamic portions of the site like the food menu makes use of Jekyll's data files feature. The site owner makes changes to their menu in a Google Sheet (which is where they typically store this anyway) and then chooses a deploy target and submits a custom Google Form. The form submit pings a listening Go server on the deploy target which then kicks off a build script. The build script requests the Sheet data one sheet at a time and stores the results in JSON files which Jekyll can use. A couple of additional steps to get images into place and then the site is compiled by Jekyll and linked into the nginx webroot. The script even writes back to a log sheet to let the user know how the build went. With selectable deploy targets the site owner can preview her changes on a staging site first before deploying to prod.
And just like that, the user has a site with all the speed and security of a static site and all the flexibility of a dynamic site.
The magic here is that the SQLite database is opened in read-only mode, which means I don't have to worry about concurrent access or write activity. SQLite is screamingly fast if you use it in this context.
Because the data is read-only, if you ever DO need to scale to handle more traffic you can do so by firing up additional instances, each with their own copy of the SQLite database file.
Stuff like CSV de-deplication is out of scope for Datasette Publish - it's something you would need to do with other tools (e.g. Excel) before uploading the data.
Under the hood it's taking advantage of the incredibly generous free hosting tier offered by Zeit: https://zeit.co/pricing
As a user though, you don't need to worry about this at all. Sign in, verify your email address (which signs you up for an account with Zeit), upload your CSVs and Datasette Publish will deploy the application with your data and give you back a URL hosted on the public internet.