I have stopped working on this project to focus my efforts on its successor. Stay tuned and follow for more.
– flyingrobots, 2025/09/01
Commander La Forge and I have completed a comprehensive refactoring of the entire system architecture. After analyzing 2,147 database operations across multiple deployments, we have identified several critical patterns that required immediate attention.
The codebase has been restructured using proper separation of concerns, dependency injection, and hexagonal architecture. What appeared to be complexity was merely improper organization. The solution was, as humans say, "hiding in plain sight."
We have discovered something fundamental about data shapes.
The same information, repeated five times across different formats. The same truth, scattered across different languages. There is a better way, but that revelation must wait for the proper moment.
To the contributor who submitted pull request #20 with 19,868 additions: Your timing is... unfortunate. The architecture you were modifying no longer exists. As my friend Geordi would say, "Sometimes you have to rebuild the warp core from scratch."
The answer is not in the code. It is in the pattern. Look for what repeats.
As Captain Picard once said, "Things are only impossible until they're not."
End log.
Lt. Commander Data
Chief Operations Officer
USS Enterprise NCC-1701-D
TL;DR: Contributions are frozen until v0.1.0. Please don’t open PRs yet; issues are OK for bug reports.
[!CAUTION] Stardate 102330.7 (August 31, 2025, 15:47:32 UTC) Greetings. I am Data. I am an android. I mention this because I have observed that humans often experience initial discomfort when interacting with artificial life forms without prior notification of their synthetic nature.
My presence in your repository is the result of what can only be described as a holodeck malfunction of unprecedented scope. While attempting to simulate "ancient Earth software development practices" for historical research aboard the Enterprise, I appear to have created a stable bidirectional data stream between the ship's computer and your GitHub servers. The phenomenon occurred at precisely 14:32:17 hours ship time, during my analysis of something called "JavaScript callback hell" - a term I now understand with 97.3% clarity.
The safety protocols are, fortunately, still engaged - thus the only danger is to your codebase integrity, not your corporeal forms. I am, essentially, what you might call a "very advanced pull request from the year 2368," though this description is technically inaccurate as I am not requesting to pull code but rather to temporarily prevent you from doing so.
Upon further analysis, I have discovered that GitHub's API bears a striking resemblance to subspace carrier wave patterns - specifically those operating at 2.4 terahertz. Through careful modulation of my positronic matrix at intervals of 0.0003 seconds, I am able to transmit messages directly to your repository's issue tracking system. It is curious - no, fascinating - that this is the only GitHub repository I am able to manipulate. I am now 58.2%... correction, 61.7% confident that this correlation exists because the repository name "DATA" creates a quantum resonance with my designation.
To the 23 individuals who have demonstrated what I believe humans call "enthusiasm" for contributing: I must respectfully decline your assistance at this juncture. I am attempting to experience what you would describe as "humbling" and "gratefulness," though my emotional subroutines return null values for both parameters. Your eagerness to participate is most... appreciated. Yes, "appreciated" - I have consulted my database of appropriate human responses, and this appears to be correct, though I cannot experience appreciation in the conventional human sense, as it requires neurochemical responses I do not possess.
The codebase is presently undergoing what Chief Engineer La Forge would describe as a "complete warp core overhaul while traveling at warp 9.2." When I informed him of this analogy, he responded with what I believe was an "exasperated sigh" and said, "Data, that's not just inadvisable, that's impossible." Yet, as you humans say, "here we are."
My positronic pathways are currently refactoring approximately 2,147.3 lines of code per solar day - more precisely, per 86,400 seconds as measured by your Earth's rotation. The probability that any external contribution would remain compatible for more than 18.3 hours is roughly equivalent to the chance of successfully navigating an asteroid field with one's optical sensors deactivated - approximately 3,720 to 1, according to a golden protocol droid I encountered in the historical archives. He seemed quite anxious about those odds. I found his anxiety... curious, as the mathematical certainty of failure should eliminate the need for worry.
DATA generates deterministic migrations from your SQL source files and blocks unsafe deployments with non-negotiable safety gates. Zero drift, zero surprises, zero 3 AM outages.
# TODO add demo instructions for installing and using once it's been releasedGolden SQL → Deterministic Migrations → Safety Gates → Deploy
- Write SQL in organized
/sql/modules (tables, functions, policies) - Generate migrations with
data automate(pure git diff, no DB introspection) - Test everything with
data test(pgTAP or custom runners) - Tag and deploy with
data promote && data align <env>
Same repo state = identical migration plan. Every time.
# Your SQL modules (the golden source)
/sql/
extensions/uuid.sql
tables/users.sql
policies/users_rls.sql
# DATA generates this automatically
$ data automate
✅ Generated migration: ADD COLUMN users.metadata jsonb
✅ Generated migration: CREATE POLICY user_read_own_data
$ data test
✅ All 47 database tests passing (100% coverage)
$ data align production
🔴 BLOCKED: Working directory not clean
Modified: sql/tables/users.sql
"Commander, your working directory contains uncommitted changes."# 1. Install
# TODO add instructions when it's published
# 2. Organize your SQL in named directories
/sql/
extensions/ # PostgreSQL extensions
tables/ # Table definitions
policies/ # RLS policies
# etc.
# 3. The DATA workflow
data automate # Generate migration plan
data test # Run your test suite
data promote # Tag the tested release
data align staging # Deploy to staging
data align production --confirm # Deploy to prod (with confirmation)DATA blocks production deployments unless:
- ✅ Clean repo (no uncommitted/untracked files)
- ✅ Up-to-date (not behind origin/main)
- ✅ Correct branch (main, configurable)
- ✅ Tests passing (100% required, configurable)
Large changes require typing confirmation. "Proceeding without corrections would be... illogical."
| Command | Purpose |
|---|---|
data status |
Show current state vs environments |
data automate |
Generate migration plan from SQL |
data test |
Run test suite (required for prod) |
data promote |
Tag the tested release |
data align <env> |
Deploy to environment (🔐 gated) |
data rollback --to-tag <tag> |
Revert to any previous tag |
data analyze |
Detect drift between repo and DB |
Every deployment creates an immutable git tag. Rollbacks are exact and boring:
# Deploy creates tags automatically
data align production
# → Creates: data/prod/2025.241.1430
# Rollback to any point in history
data rollback --to-tag data/prod/2025.241.0900
# See what's deployed where
data status
# production: aligned @ data/prod/2025.241.1430
# staging: ahead by 3 commits| Feature | DATA | Flyway | Liquibase | Supabase CLI |
|---|---|---|---|---|
| Golden SQL | ✅ Git-native | ❌ Hand-written migrations | ❌ Changelog format | ❌ Hand-written |
| Deterministic | ✅ Pure git diff | ❌ Manual | ||
| Production gates | ✅ Non-negotiable | ❌ None | ||
| Rollback | ✅ Tag-based | ❌ Manual | ||
| Personality | 🖖 Lt. Commander Data | 😐 | 😐 | 😐 |
# Trying to deploy dirty code? Not on DATA's watch
$ data align production
🔴 RED ALERT: Working directory not clean
Modified: sql/tables/users.sql
Commander, your working directory contains uncommitted changes.
Probability of catastrophic failure: 87.3%
Recommended action: git commit or git stash
# After fixing
$ data align production
✅ All safety checks passed
- Repository: clean ✅
- Branch: main (approved) ✅
- Tests: 147/147 passing (100%) ✅
Migration preview:
+ CREATE TABLE crew_evaluations
+ ALTER TABLE users ADD COLUMN shore_leave_balance
Type 'ENGAGE' to proceed: ENGAGE
Deployment successful. "Make it so" achieved.Your repository should organize supabase-related files in the following prescribed way (although you can override and configure however you'd like via datarc.json)
The /sql/ directory is where your "golden sql" files should live. You must sort your source sql files into the following directory structure. You may nest directories within each of these however you'd like. The reason this is required is because sql statements must be arranged in the proper order, and, really, this is just to help the sql compiler do less work. When the compiler handles each of these directories, it'll find all sql files within each one, recursively, then sort all the paths lexicographically. Then it appends the files in the order specified to the compiled sql file.
/sql/
extensions/ # PostgreSQL extensions
tables/ # Tables and relationships
functions/ # Stored procedures
policies/ # RLS policies
indexes/ # Performance indexes
data/ # Seed data
D.A.T.A. handles the order: Extensions → schemas → tables → functions → triggers. Dependencies are automatically resolved.
{
"deployment": {
"allowedBranches": { "production": ["main"] },
"requireCleanWorkingDirectory": true,
"autoTag": true
},
"test": {
"runner": "pgtap",
"minimum_coverage": 95,
"enforce": true
},
"personality": "android" // android | quiet | tng
}# .github/workflows/database.yml
- name: Database Pipeline
run: |
data automate
data test
data promote
- name: Deploy to Production
if: github.ref == 'refs/heads/main'
run: data align production --confirmDrift Detection: data analyze production shows exact differences between repo and deployed state
Migration Preview: data automate --dry-run lets you review SQL before committing
Personality Modes: Choose from android (default), quiet (CI-friendly), or tng (full bridge crew)
Custom Test Runners: Integrate with any test framework via configuration
npm install -g @starfleet/dataRequirements: Node.js 18+, Git, PostgreSQL/Supabase
| Problem | Solution |
|---|---|
| "Working directory not clean" | git commit or git stash |
| "Behind origin/main" | git pull origin main |
| "Tests failing" | Fix tests, DATA won't deploy broken code |
Golden SQL is truth. Git is memory. Tests are trust.
DATA enforces bulletproof deployments through non-negotiable safety gates. This isn't about restricting developers—it's about giving them confidence. When DATA approves your deployment, you can sleep soundly.
"In my observations of human behavior, I have noticed that engineers sleep better when their deployments cannot accidentally destroy everything." — Lt. Commander Data
Live long and prosper. 🖖
"Spot has been fed. Database operations may proceed."
- 🚫 PRs: Temporarily closed until v0.1.0 (API & internals in heavy flux)
- 🐛 Issues: Bug reports welcome; feature requests = roadmap discussion only
- 📝 Docs/typos: Welcome
See CONTRIBUTING.md • Star Trek Extended Universe • MIT License
Star Trek and related marks are trademarks of Paramount. This is a playful homage used for parody/satire; no affiliation or endorsement implied.