frontend-patterns/research/candidates/VerticalFeatureSlice.md
tegwick 50c68575f1 Broaden pattern hypotheses across presentation composition and quality
Assistant: codex
Assistant-Model: gpt-6-astra
Assistant-Session: 01a0cffe-206f-7323-89d3-b20ca4cf4bf9
2026-09-26 21:46:28 +02:00

40 lines
1.9 KiB
Markdown

---
title: Vertical Feature Slice
status: draft
maturity: hypothesis
families: [composition]
scales: [L4]
admission: research-candidate-only
---
# Vertical Feature Slice
Context: one product change spans UI, state and service interaction scattered across
technical folders. Change locality competes with reuse and cross-feature coordination.
Proposed solution: group the frontend parts of a cohesive capability behind an
explicit public interface. Declare permitted dependencies and compose capabilities
at a higher boundary. Put shared primitives where their ownership is clear; do not
hide cross-feature workflows inside another feature's private modules.
Result: a capability can evolve internally, but interface design and shared concepts
need maintenance. A folder or barrel export does not enforce isolation. This is a
source-code boundary, not proof of independent deployment, team autonomy, security
isolation or an end-to-end backend service boundary.
Evidence: B03 locates Feature-Sliced Design's product-oriented grouping and public
API rules. The proposed pattern abstracts those mechanisms without adopting its
entire layer taxonomy. Full-pattern application recurrence remains unverified.
Alternative: technical layers for a small cohesive application; extract shared
behavior only when its ownership and consumers justify the new coupling.
Proposed relation: may contain [Presentation Model](PresentationModel.md); feature
boundaries need not coincide with [Failure Boundary](FailureBoundary.md).
Next test: change one capability's storage adapter; inspect imports and consumer
contracts. Change only deployment cadence next and identify new obligations.
Evidence locators: [breadth review](../reviews/PatternBreadthExpansion.md).
All relations are research proposals; no canonical identity or admitted graph edges.
The next test is planned, not executed. Known uses of the complete proposed contract
have not been verified in this investigation.