frontend-patterns/research/candidates/PresentationModel.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

39 lines
1.8 KiB
Markdown

---
title: Presentation Model
status: draft
maturity: hypothesis
families: [composition, affordance]
scales: [L4, L6]
admission: research-candidate-only
---
# Presentation Model
Context: a screen contains presentation decisions that are difficult to exercise
through widget APIs. Testability and reuse compete with binding complexity.
Proposed solution: represent presentation state and decisions independently of the
concrete controls; adapt that model into the view and translate interaction back
into intents. Domain authority stays outside presentation decisions. Specify the
binding direction and ownership of each editable value to avoid feedback loops.
Result: logic can be tested without mounting the full interface, but model/view
synchronization becomes another failure surface. This is broader than Local Draft
State: a presentation model may contain selection and enablement without any draft.
Do not create a second editable copy of every domain object by default.
Evidence: B01 locates Fowler's named pattern and its synchronization trade-off.
The source is a historical draft, not a verified deployment in this research.
Alternative: keep a trivial screen's logic in its view; another controller style
may be sufficient. Full-pattern recurrence here remains unverified.
Proposed relations: may contain [Local Draft State](LocalDraftState.md) and compute
[Derived State](DerivedState.md); neither is mandatory.
Next test: edit, refresh and change selection; verify ownership of draft versus
server observation and ensure a binding update does not trigger duplicate commands.
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.