Datomic Classes

Companion materials for two classes:

• One Database, Many Shapes — a 2-hour, REPL-first class on domain modeling with Datomic datoms.
• Datomic in Production — four 2-hour, hands-on sessions on Datomic as infrastructure: deployment topologies, the read and write paths, storage and persistence modes, the physical cache tiers, and admin operations. Labs run Datomic Pro against PostgreSQL in Docker; Datomic Cloud and Datomic Local are covered by comparison.

Files

One Database, Many Shapes

• domain-modeling-with-datoms.pptx - the slide deck.
• build_deck.py + deck_content.py - the deck's source. python3 build_deck.py regenerates the pptx.
• src/domain_modeling/repl.clj - the live REPL companion. Slide anchors in the file match the deck's REPL §n slides.
• src/domain_modeling/time_exercises.clj - "fill the gaps" exercises for the last 30 minutes, built on Part 4 (Time): tx provenance, cell-level blame, reverting a transaction, and schema evolution under a full audit. Eval (domain-modeling.time-exercises/start!) to begin; solutions at the bottom of the file.

Datomic in Production

• datomic-infrastructure.pptx - the 4-session slide deck.
• build_infra_deck.py + infra_content.py + infra_part1..4.py - its source, one module per session. python3 build_infra_deck.py regenerates the pptx and fails the build if a slide's content overflows the canvas.
• infra_layout.py - the flow-layout DSL the sessions are authored in (BUL, ROW, CARDS, TABLE, CODE, NOTE, TWO), so slides are written as content rather than as EMU coordinates.
• src/datomic_infra/labs.clj - the lab companion; sections match the deck's sessions and every ◆ lab slide.
• infra/ - docker-compose.yml (PostgreSQL, plus memcached for session 3), a transactor properties file, and setup instructions.

Shared

• deck_shell.py + deck_builder.py - pptx packaging for both decks: plain text boxes on a Keynote-authored master/layout/theme, so the files import cleanly in macOS Keynote (the original Walnut export did not).
• deps.edn - Clojure CLI dependencies. The :infra alias adds the PostgreSQL JDBC driver the peer needs to reach SQL storage.

Prerequisites

• JDK 11 or newer.
• Clojure CLI tools.

One Database, Many Shapes runs entirely in memory with datomic:mem://, so no Datomic transactor, storage service, account, or external database is required.

Datomic in Production is about the infrastructure, so it needs the real thing: Docker (for PostgreSQL and memcached) and a Datomic Pro distribution unzipped locally, for bin/transactor and bin/datomic. Datomic Pro has been Apache-2.0 licensed and free since 2023 — no account, no license key. See infra/README.md.

Start The REPL

Clone the repo and start the interactive nREPL:

git clone https://github.com/4coders-com-br/datomic-model.git
cd datomic-model
clj -M:repl

From the REPL, load the companion namespace:

(require 'domain-modeling.repl)

The namespace creates an in-memory Datomic connection on load. Nothing from the class blocks runs automatically; the examples live inside (comment ...) forms so you can evaluate them one expression at a time.

Follow The Deck

Open domain-modeling-with-datoms.pptx, then use the SLIDE n anchors in src/domain_modeling/repl.clj.

If you jump into the middle of the class, evaluate the matching goto! call under that slide anchor first:

(domain-modeling.repl/goto! 24)

goto! rebuilds the in-memory database from scratch to the exact starting state for that slide, then prints the current class section and basis-t. For the exercises (slide 41, end of Part 5):

(domain-modeling.repl/goto! 41)

Expected result:

[:slide 41 :ready]

Reset Manually

To clear the in-memory database at any time:

(domain-modeling.repl/fresh!)

Then continue top to bottom, or use goto! to rebuild the state for a specific slide.