docs: fix type preservation claims in comparison.md

[27Bslash6]

Summary

Update comparison.md to reflect that L1-only mode (backend=None) now preserves Python types after #73/#117.

Three claims corrected:

• Feature comparison note: tuples/sets preserved in L1-only, only converted with distributed backends
• Single-process section: removed incorrect "tradeoff" note about serialization
• Validation evidence: updated to reflect L1-only stores raw objects

Closes #74

Summary by CodeRabbit

• Documentation
  • Clarified type preservation differences between single-process (L1-only) and distributed cache modes.
  • Stated single-process mode stores raw Python objects and avoids serialization overhead.
  • Noted distributed backends perform serialization that can convert tuples/frozensets to lists by default.
  • Added guidance on configuring serialization to preserve Python types with distributed backends.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro

Run ID: c51152bd-fd87-494d-9241-615c23ac25ff

📥 Commits

Reviewing files that changed from the base of the PR and between cac53a1 and 7d56dc2.

📒 Files selected for processing (1)

• docs/comparison.md

🚧 Files skipped from review as they are similar to previous changes (1)

• docs/comparison.md

---

📝 Walkthrough

Walkthrough

This PR updates docs/comparison.md to correct claims about type preservation: L1-only (backend=None) preserves raw Python types without serialization, while distributed backends use MessagePack (StandardSerializer) which converts tuples/frozensets to lists by default unless serializer='auto' is configured. Three related sections are updated.

Changes

Type Preservation Documentation

Layer / File(s)

Summary

Type preservation clarification across three sections docs/comparison.md

Updated the "Type preservation" guidance to distinguish L1-only (backend=None) raw-object storage from distributed-backend serialization (MessagePack StandardSerializer converting tuples/frozensets to lists), replaced the single-process "tradeoff" text with a "Type safe" note about L1-only, and adjusted the validation evidence bullet to state L1-only preserves Python types while distributed backends serialize unless serializer='auto' is used.

Estimated code review effort

🎯 1 (Trivial) | ⏱️ ~3 minutes

Possibly related PRs

• cachekit-io/cachekit-py#117: Implements the code-level change that preserves raw Python types in L1-only mode while distributed backends serialize, aligning docs with behavior.

Poem

🐰 I hopped through lines of docs so neat,

Swapped a claim for clarity, tidy and sweet.
L1 keeps objects whole, no serialize art—
Distant backends pack and convert apart.
A small carrot dance for docfix complete.

🚥 Pre-merge checks | ✅ 3 | ❌ 2

❌ Failed checks (1 warning, 1 inconclusive)

Check name	Status	Explanation	Resolution
Description check	⚠️ Warning	The PR description does not follow the required template structure and is missing required sections like Type of Change, Security Checklist, and Documentation Validation Checklist.	Restructure the description to match the template, including the Type of Change checkbox, Security and Documentation validation checklists, and Testing sections.
Linked Issues check	❓ Inconclusive	The PR claims to fix type preservation claims but the implementation appears incomplete relative to the issue's suggested fix requirements.	Clarify whether the changes fully address issue #74's suggestion that cachekit serializes in ALL modes, or only partially address the specific type preservation claims.

✅ Passed checks (3 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title accurately summarizes the main change: correcting type preservation claims in the comparison documentation.
Out of Scope Changes check	✅ Passed	All changes in the PR are scoped to updating documentation claims in comparison.md to accurately reflect cachekit's actual type handling behavior, with no unrelated modifications.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch docs/fix-comparison-type-claims

---

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ All tests successful. No failed tests found.

📢 Thoughts on this report? Let us know!