spec-proposal

name	spec-proposal
description	Write a feature spec or bug analysis for a Tensor4all.jl issue

Spec Proposal

You are helping a contributor write a spec for a Tensor4all.jl GitHub issue.

Before you start

1. Read CONTRIBUTING.md for the contribution flow.
2. Read AGENTS.md for codebase conventions and architecture.
3. Understand the issue: what is being proposed or reported.

Your task

Explore the codebase to understand the relevant modules, types, and existing behavior. Then draft a spec to post as the issue description or comment.

Output format

Write the spec with these sections:

## Summary
(What to add or change, 1-2 sentences)

## Motivation
(Why this is needed, which use case)

## Proposed approach
(Recommended design direction. Reference relevant modules/types.
 Pseudocode is fine; exact code is not required.)

## Alternatives considered
(Other approaches and why they were not chosen)

## Acceptance criteria
- [ ] ...
- [ ] ...

Guidelines

• Keep it concise. Focus on the "what" and "why", not the "how" in detail.
• Reference module names (e.g. TensorNetworks, SimpleTT) but avoid hardcoding line numbers — they change.
• If the change touches the C API boundary, note that explicitly.
• Validate that the proposed approach is consistent with AGENTS.md design principles (e.g. Julia-owned semantics, minimized C API).
• If the feature requires new C API functions in tensor4all-rs, state this explicitly and suggest opening a linked issue on the tensor4all-rs repository. See the "Cross-Repository Changes" section of CONTRIBUTING.md.