research

Research Notes

Status: exploratory index

This directory contains design notes, open questions, architectural explorations, and long-horizon ideas for Concrete.

These files are exploratory unless they explicitly say otherwise. Once a design becomes a stable project rule or implementation contract, it should move into docs/.

How To Use This Directory

• use ROADMAP.md for active project sequencing
• use docs/ for stable rules and implementation contracts
• use research/ for design work that is still being explored, sharpened, or staged for later phases

Directory Layout

• research/language/ — language-surface decisions, exclusions, and evidence-gated language ideas
• research/stdlib-runtime/ — stdlib shape, text/string direction, collections, execution/runtime pressure, targets, and resource models
• research/compiler/ — compiler architecture, backend direction, artifact/dataflow work, semantic diffing, and performance/backend research
• research/proof-evidence/ — formalization, proof workflows, evidence artifacts, trust bundles, and review/report direction
• research/packages-tooling/ — package model, testing workflow, developer tooling, authority budgets, and project/workspace UX
• research/workloads/ — Phase H findings, comparison suites, showcase workloads, and example-driven pressure notes
• research/process/ — quality/process notes such as testing strategy
• research/meta/ — broader strategy, gap analysis, candidate ideas, and long-horizon synthesis

Priority Key

• P0 = highest-value current research, directly connected to active or next roadmap phases
• P1 = important follow-on research, likely to matter in later phases
• P2 = useful background, optional direction, or long-horizon exploration

Status Key

• Open = still exploratory
• Adopted = design influenced implementation, but the note remains useful as background
• Excluded = intentionally not in the language surface
• Process = decision filter or project rule
• Research = broader exploration, not a current design commitment
• Closed = the design question is no longer open, but the note remains as historical context

Start Here

If you want the highest-leverage current research first:

1. workloads/phase-h-findings.md — classified findings from the real-program corpus (P0, Open)
2. proof-evidence/formalization-roi.md — what to prove first and why (P0, Open)
3. proof-evidence/formalization-breakdown.md — the full formalization effort split into tracks and milestones (P0, Open)
4. compiler/artifact-driven-compiler.md — stable artifacts, IDs, traceability, and the real compiler driver (P0, Open)
5. packages-tooling/package-model.md — the eventual package/dependency model (P0, Open)
6. packages-tooling/package-testing-tooling.md — package/workspace test workflow design (P0, Open)
7. proof-evidence/evidence-review-workflows.md — evidence bundles, machine-readable reports, and review workflows (P0, Open)
8. compiler/semantic-diff-and-trust-drift.md — semantic/package/release diffing over compiler facts (P0, Open)
9. stdlib-runtime/text-and-output-design.md — current text/output direction (P0, Open)
10. stdlib-runtime/runtime-collections.md — collection maturity for interpreter/runtime workloads (P1, Open)
11. stdlib-runtime/iterators.md — explicit traversal support without an iterator tower (P1, Open)
12. meta/high-leverage-systems-ideas.md — recurring high-value systems ideas (P0, Open)
13. proof-evidence/trust-multipliers.md — the strongest combined differentiators across proof, runtime, and evidence (P0, Open)
14. compiler/qbe-backend.md — lightweight alternate-backend research (P1, Open)
15. meta/ai-assisted-optimization.md — optimization/refactoring loops driven by structured compiler reports (P1, Research)

Language

• language/builtin-vs-stdlib.md — what belongs in compiler/runtime builtins versus the public stdlib (P0, Open, partially adopted)
• language/capability-sandboxing.md — restricted authority and sandboxing direction (P0, Open, partially adopted)
• language/high-integrity-profile.md — stricter profile/subset for critical code (P0, Open)
• language/binary-format-dsl.md — binary parser/serializer DSL direction (P2, Research)
• language/cleanup-ergonomics.md — reducing ownership/cleanup friction without hiding it (P1, Open)
• language/const-generics-comptime.md — const generics versus comptime, and where the philosophy line should stay (P2, Research)
• language/derived-equality-design.md — possible derived structural equality (P2, Open)
• language/heap-ownership-design.md — chosen Heap<T> ownership model (P1, Adopted)
• language/heap-access-revisited.md — follow-up on heap access syntax and tradeoffs (P2, Open)
• language/module-qualification.md — historical note on qualified access and namespace pressure (P1, Closed)
• language/no-closures.md — why Concrete excludes closures (P1, Excluded)
• language/no-trait-objects.md — why Concrete excludes trait objects (P1, Excluded)
• language/pre-post-conditions.md — contracts/specification support and why it stays later/optional (P1, Open)
• language/trusted-boundary.md — explicit trusted fn / trusted impl containment design (P1, Adopted)
• language/typestate.md — typestate direction and limits (P1, Open)
• language/union.md — whether unions fit Concrete’s design (P2, Open)
• language/unsafe-structure.md — making Unsafe more inspectable without a heavier language (P1, Open, partially adopted)

Stdlib And Runtime

• stdlib-runtime/stdlib-design.md — stdlib direction, priorities, and style rules (P1, Open, partially adopted)
• stdlib-runtime/stdlib-api-cleanup.md — cleaning builtin-shaped names and ownership surprises out of the public stdlib (P1, Open)
• stdlib-runtime/text-and-output-design.md — mixed-arg printing first, interpolation later if needed (P0, Open)
• stdlib-runtime/runtime-collections.md — collection maturity for interpreter/runtime workloads (P1, Open)
• stdlib-runtime/iterators.md — explicit traversal support and the case against a broad iterator ecosystem (P1, Open)
• stdlib-runtime/arena-allocation.md — arena/bump allocation for parser/interpreter-style workloads (P1, Open)
• stdlib-runtime/allocation-budgets.md — NoAlloc, BoundedAlloc(N), and allocation reporting (P0, Open)
• stdlib-runtime/execution-cost.md — structural cost reports and bounded execution direction (P1, Open)
• stdlib-runtime/execution-cost-tracking.md — tracking/externalizing execution-cost obligations (P1, Open)
• stdlib-runtime/layout-reports.md — layout/ABI audit report improvements (P1, Open)
• stdlib-runtime/concurrency.md — near-term concurrency direction (P1, Open)
• stdlib-runtime/long-term-concurrency.md — long-horizon layered concurrency model (P1, Open)
• stdlib-runtime/no-std-freestanding.md — hosted vs freestanding / no_std split (P1, Open)
• stdlib-runtime/runtime-execution-pressure.md — runtime/stack pressure from deep-recursive workloads (P1, Open)
• stdlib-runtime/target-platform-policy.md — support tiers, ABI promises, and target policy (P1, Open)
• stdlib-runtime/hardware-capability-mapping.md — hardware-backed capability ideas (P2, Research)

Compiler

• compiler/artifact-driven-compiler.md — operationalizing named compiler artifacts (P0, Open)
• compiler/compiler-dataflow-ideas.md — dataflow/query-inspired artifact and inspection ideas (P1, Open)
• compiler/external-ll1-checker.md — external grammar + LL(1) checker (P1, Open)
• compiler/file-summary-frontend.md — summary-based frontend direction (P1, Adopted)
• compiler/ll1-grammar.md — strict LL(1) rule and parser cleanup criteria (P1, Process)
• compiler/mlir-backend-shape.md — where MLIR should sit if it earns its complexity (P1, Research)
• compiler/miri-style-interpreter.md — interpreter/tooling direction for stronger semantic checking (P2, Research)
• compiler/optimization-policy.md — optimization goals, non-goals, and observability constraints (P1, Open)
• compiler/qbe-backend.md — lightweight alternate-backend research (P1, Open)
• compiler/qbe-in-concrete.md — long-horizon self-hosting/backend-implementation idea (P2, Research)
• compiler/semantic-diff-and-trust-drift.md — semantic/trust-drift diffing over compiler facts (P0, Open)
• compiler/semantic-diff.md — broader semantic diff/query ideas (P2, Research)
• compiler/vec-inline-investigation.md — VM/codegen inline-cliff investigation (P1, Adopted)

Proof And Evidence

• proof-evidence/formalization-roi.md — best order for proving semantics and guarantees (P0, Open)
• proof-evidence/formalization-breakdown.md — full proof effort split into tracks and dependencies (P0, Open)
• proof-evidence/proving-concrete-functions-in-lean.md — how selected Concrete functions could be proved in Lean 4 (P0, Open)
• proof-evidence/proof-addon-architecture.md — proof automation as an artifact-consuming addon workflow (P0, Open)
• proof-evidence/proof-evidence-artifacts.md — tying reports, artifacts, proofs, and reproducibility together (P0, Open)
• proof-evidence/evidence-review-workflows.md — evidence bundles, review workflows, and maintained reports (P0, Open)
• proof-evidence/trust-multipliers.md — combined differentiators across proof, runtime, and evidence (P0, Open)
• proof-evidence/cryptographic-source-commitments.md — cryptographic source/bundle provenance (P1, Open)
• proof-evidence/proof-carrying-supply-chain.md — proof/evidence-aware supply-chain direction (P2, Research)
• proof-evidence/self-describing-binaries.md — binary-to-proof/evidence traceability direction (P2, Research)

Packages And Tooling

• packages-tooling/package-model.md — package identity, dependency semantics, and workspaces (P0, Open)
• packages-tooling/package-manager-design.md — CLI/manifest/lockfile/graph shape (P0, Open)
• packages-tooling/package-testing-tooling.md — package-aware concrete test workflow (P0, Open)
• packages-tooling/authority-budgets.md — authority budgets and dependency policy (P0, Open)
• packages-tooling/developer-tooling.md — editor/LSP, debugging, and CLI workflow direction (P1, Open)
• packages-tooling/repl-and-playground.md — REPL/playground direction (P2, Research)
• packages-tooling/standalone-vs-project-ux.md — single-file versus project UX (P1, Open)

Workloads And Examples

• workloads/phase-h-findings.md — classified findings from real programs (P0, Open)
• workloads/phase-h-summary.md — canonical summary of what Phase H taught (P0, Open)
• workloads/comparative-program-suite.md — real-program comparison portfolio against Rust/Zig/C (P1, Open)
• workloads/showcase-workloads.md — serious programs Concrete should eventually implement well (P1, Open)
• workloads/high-integrity-examples.md — example shapes for the future high-integrity profile (P1, Open)
• workloads/adoption-strategy.md — signature domains, showcases, onboarding, and public pull (P1, Open)

Process And Quality

• process/testing-strategy.md — gaps beyond the current suites: fuzzing, property tests, and differential testing (P1, Open)

Meta And Long-Horizon

• meta/high-leverage-systems-ideas.md — recurring high-value systems ideas (P0, Open)
• meta/ten-x-improvements.md — the biggest long-term multipliers for Concrete (P0, Open)
• meta/competitive-gap-analysis.md — what other systems languages still have and which gaps matter (P1, Open)
• meta/complete-language-system.md — what still separates a strong compiler from a complete language system (P1, Open)
• meta/ai-assisted-optimization.md — optimization/refactoring guided by structured report outputs (P1, Research)
• meta/candidate-ideas.md — Concrete-specific candidate language/compiler/tooling ideas (P2, Research)
• meta/external-ideas.md — useful ideas borrowed from other languages (P2, Research)

Placement Rule

• stable rule/reference -> docs/
• active plan/sequencing -> ROADMAP.md
• landed milestone/history -> CHANGELOG.md
• exploratory note -> research/

The roadmap should only absorb items from here when they become concrete technical work.