Under the hood

How TableSpark works, in depth

Feature lists all look the same. What separates restaurant software is what it does on a bad night: two guests on one table, a payment that half went through, a guest who asks to be forgotten.

228 statements across six pages, each traced to code or a test.

Why these pages exist

A restaurant choosing software is shown lists of features, and every list looks alike. The questions that decide whether a service goes well are not on the list. What happens when two guests book the last table at once? When a card reader times out halfway through a payment? When a guest asks to be forgotten? When something breaks at seven on a Saturday?

A wrong answer to any of them is paid for in the room: a party turned away at the door, a till that does not match the bank, a complaint to the regulator, an evening of lost orders. And an owner comparing products has no way to tell a careful answer from a confident one.

So these pages give the answers in a form that can be checked. Each takes one question, starts from the restaurant’s side of it, and then sets out the mechanism precisely enough for an engineer to hold us to it.

Six questions a careful buyer asks

How these pages are kept true

The six pages make 228 statements about how TableSpark behaves. None of them is typed into a page. Each is held in one file together with its evidence: the place in the code or the database that enforces it, and the automated suite that exercises it. The pages are generated from that file.

  • An automated check fails when a file or a test named as evidence no longer exists, or no longer contains what the statement relies on.
  • Only behaviour that is live in production is described.
  • Counts, such as the number of automated test files, are read from the repository when the pages are built.

More to check

See it on your own restaurant

Build your site free and publish when you are ready. Plans start at £19 a month, excluding VAT.

Start building free