Roadmap
What Quantum is today, and what comes next. There are no dates here, and no promises beyond the support tiers: an item is done when it has shipped in a release and its changelog entry says so.
Each item shows its status: planned (approved, not started), in design (being written down before any code) or in progress (being built now).
1.0 — today
Quantum 1.0 is declarative web applications in XML, with AI and RAG built into the language. What 1.0 promises is set by its tiers:
- Core (components, queries and transactions, actions and forms, files, mail, authentication, the Core set of
ui:*) and AI (q:llm,q:knowledge,q:agent) follow semantic versioning: a 1.x release does not break a program that uses only them. AI is also tested against a real model before every release. - Experimental tags work, with no stability promise.
- Laboratory projects stay in the repository because they push on the language; they are not part of the product and carry no promise.
What runs today is measured, not claimed: see Status. Every tag and rule is in the Reference.
Next
AI
| Item | Status |
|---|---|
| A fake model for development: build and test AI pages with no model running, with answers that are the same on every run | planned |
| AI in the dev panel: every call a page made, with its prompt, the sources it used, the answer, the time and the tokens | planned |
| Budgets and redaction: a spending limit per application, and personal data removed before it leaves for the model, with the record of both kept | planned |
Conversations that remember, and a ui:chat to hold them | planned |
An application that knows itself
| Item | Status |
|---|---|
quantum map: which pages and actions read and write each table | planned |
quantum explain page.q: what a page reads, writes and depends on, in plain text | planned |
q:cache by table: a cached fragment that is refreshed when a write touches one of its tables, with no expiry time to guess | planned |
Testing
quantum test shipped in 1.0 (Testing). Next:
| Item | Status |
|---|---|
| One test run against every screen: the browser and the console | planned |
| Tests derived from an action's rules, written for you | planned |
| AI tests replayed from a recording, so they run without a model | planned |
| Coverage in the application's own terms: which pages, actions and rules a suite reaches | planned |
The website
| Item | Status |
|---|---|
| Cookbook: short recipes, each one tested in CI | in progress |
| The site in more languages | in progress |
| A playground: write Quantum and run it in the browser | planned |
| Benchmarks that anyone can run again | planned |
Laboratory
| Item | Status |
|---|---|
| Games 2: games as a deterministic declarative simulation, where the same inputs always give the same game, so a game can be tested and replayed | in design |
Laboratory work carries no promise: it can change or stop at any point.