Skip to content

docs(tdd): add TDD for 1.54 QA follow-up security and stability fixes - #18584

Merged
christian-byrne merged 4 commits into
mainfrom
docs/qa-followup-tdd
Sep 27, 2026
Merged

christian-byrne merged 4 commits into
mainfrom
docs/qa-followup-tdd

Conversation

@claude

@claude claude Bot commented Sep 23, 2026

Copy link
Copy Markdown
Contributor

Requested by Christian Byrne · Slack thread

Summary

Before: the six remaining bugs from the 1.54 QA sweep (three sanitization/XSS gaps, a group-node widget-pairing bug, a VirtualGrid blank-render bug, and a preview-widget text-stripping bug) were being fixed as four independent PRs with no document tying them together as one scope contract — no shared record of root cause, fix approach, or verification plan across the batch.

After: a single Technical Design Document (docs/tdd/2026-09-qa-followup-security-and-stability-fixes.md) covers all six items: problem statement and severity, exact scope (which PR/branch owns which fix), a per-item design section with root cause and fix approach, a sequence diagram for the VirtualGrid clamp-ordering bug, a verification plan (one row per acceptance criterion), and a rollout section noting these are plain merges with no flag needed, with the core/1.54 backport decision explicitly left to the release owner. This is a docs-only PR; no application code changes.

How

Grounded the write-up by reading the actual source for each of the six fixes as they stand on main (markdownRendererUtil.ts, SanitizedHtml.vue, DescriptionTabPanel.vue, groupNode.ts, VirtualGrid.vue, TextPreviewWidget.vue), and by reading the two existing pinned-test PRs (#17464 for the group-node case, #17626 for the VirtualGrid case) that document those two defects in detail. For the VirtualGrid fix, also read the in-progress diff already drafted on branch fix/virtual-grid-clamp-blank to confirm the fix approach described matches what's actually being built. The other three fix branches (fix/sanitization-hardening, fix/group-node-widget-pairing, fix/preview-widget-bracket-text) had no commits yet at the time of writing, so their fix-approach sections describe the intended approach rather than an already-implemented diff. Followed the comfy-tdd skill's document skeleton (contract, design, verification, rollout), adapted for a bug-fix batch: no PRD (none exists for a QA-sweep-originated fix), no flags/ramp (each fix corrects already-broken behavior rather than adding new gated behavior), and DRI/approver/post-merge-owner left as open questions in Part 5 rather than invented.


Generated by Claude Code

Adds a shared Technical Design Document covering the scope, root cause,
fix approach, and verification plan for the six 1.54 QA-sweep bugs
being fixed across four parallel PRs: sanitization hardening,
group-node widget-pairing, VirtualGrid clamp-ordering, and the
preview-widget bracket-text stripping bug. Docs only, no code changes.
@coderabbitai

coderabbitai Bot commented Sep 23, 2026 •

Copy link
Copy Markdown
Contributor

Important

Review skipped

Bot user detected.

To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Repository: Comfy-Org/ComfyUI_frontend/.coderabbit.yaml

Review profile: ASSERTIVE

Plan: Advanced

Run ID: 3ce226c6-4050-4442-ac93-73b2618d66c1

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

  • 🔍 Trigger review

Comment @coderabbitai help to get the list of available commands.

@claude
claude Bot requested a review from a team September 23, 2026 10:07
@github-actions

github-actions Bot commented Sep 23, 2026 •

Copy link
Copy Markdown
Contributor

🧩 Custom Nodes Ecosystem Matrix: 🟡 Withheld — could not measure

Executes the frontend JS of every registry pack that ships any, against this change, in both renderers. The eight ecosystem-matrix shard checks are a harness self-test; the pack verdict is matrix-verdict.

This candidate changes nothing the matrix measures, so every expensive stage was skipped.

Reproduce locally
# build one shard's fixture, then run it
python3 scripts/registry-census/build_matrix.py --shard 1/4
MATRIX_RENDERER=vue MATRIX_VUE=1 \
  npx vitest run --config vitest.matrix.config.mts

Full docs: scripts/registry-census/README.md

🌾 Fallow: ✅ Passed

Fallow audits the diff for duplication, complexity and dead code. It gates on new findings only, so when the audit completes, a red is something this PR introduced or moved — never inherited backlog. A red can also mean the audit itself did not finish; the details below say which happened.

No new findings in the changed files.

Reproduce locally
# the base the gate compared against
npx fallow audit --root . --quiet --format json \
  --changed-since 6afd50d7a93f607a6c0bd6504d4ff06a96d07e10

Full docs: ADR-DEVEX-LINT-0015

@github-actions

Copy link
Copy Markdown
Contributor

🔧 Auto-fixes Applied

This PR has been automatically updated to fix linting and formatting issues.

⚠️ Important: Your local branch is now behind. Run git pull before making additional changes to avoid conflicts.

Changes made:

  • ESLint auto-fixes
  • Oxfmt formatting

@github-actions github-actions Bot added the risk:medium PR risk grade (advisory shadow check; grader-owned) label Sep 23, 2026
DrJKL
DrJKL previously requested changes Sep 26, 2026

@DrJKL DrJKL left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Three design-contract blockers remain:

  • AC3 describes an unreachable direct-license URL path and does not define canonicalization or the actual generated-license-link sink.
  • The VirtualGrid algorithm and provenance do not match the referenced branch implementation.
  • The planned group-node gate never asserts the two colliding widget values.

I also found four material or minor documentation/test-plan issues inline. These should be addressed before downstream fix PRs use this TDD as their contract.

Checked and cleared
  • Keeping one shared TDD is an explicit workstream decision, not accidental duplication.
  • Sanitization hardening is consistently treated as one of four independently mergeable PR artifacts.
  • Missing post-merge ownership and abort criteria are explicitly recorded draft gaps, not hidden omissions.
  • /no-comments found one redundant Mermaid comment. I removed it in ca23aeb3c; no other scoped comments should be deleted.
Coverage and verification

Standard review covered correctness, maintainability, verification quality, documentation, and /no-comments. I traced the six proposed fixes against the named source seams and PRs #17464/#17626, reproduced the markdown breakout and URL-canonicalization test gaps, and compared the VirtualGrid section with its branch implementation. git diff --check and scoped oxfmt --check pass on the final head. Exact-head CI was checked before submission.

@github-actions github-actions Bot added risk:high PR risk grade (advisory shadow check; grader-owned) and removed risk:medium PR risk grade (advisory shadow check; grader-owned) labels Sep 26, 2026
Rewrites AC3 around the two real DescriptionTabPanel.vue anchor sinks
(safeRepositoryHref and safeLicenseHref) and specifies concrete
scheme-canonicalization test cases, since direct license text is
never bound to href. Updates the VirtualGrid design, fix approach,
and sequence diagram to match the actual clamp-end-first-then-start
algorithm landed on fix/virtual-grid-clamp-blank, citing the immutable
commit instead of an in-progress draft. Requires the group-node
verification plan to assert both colliding CLIPTextEncode.text values
by stable node identity, not just downstream KSampler/SaveImage
values. Also specifies a real pre-fix falsifier for the markdown-title
sanitizer test, narrows the no-flag rationale to "restores intended
behavior, adds no new feature" instead of claiming no visible-output
change, relabels the markdown-attribute-breakout item as HTML/form/
UI-redress injection rather than XSS (DOMPurify's final pass strips
event handlers regardless of that bug), and replaces the unsupported
"R1 = small reversible diffs" claim about .github/risk.json with the
file's actual path-floor/provenance/reversibility grading mechanism.
@claude
claude Bot force-pushed the docs/qa-followup-tdd branch from a1a2a0e to 5ae13c5 Compare September 26, 2026 03:11
@github-actions github-actions Bot added risk:medium PR risk grade (advisory shadow check; grader-owned) and removed risk:high PR risk grade (advisory shadow check; grader-owned) labels Sep 26, 2026

@christian-byrne christian-byrne left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Re-reviewed the current head against all seven prior findings and the implementation PRs. The TDD now matches the shipped algorithms, security classification, verification requirements, and no-flag rationale.

@christian-byrne
christian-byrne dismissed DrJKL’s stale review September 27, 2026 04:53

All seven findings were addressed in follow-up commits and independently re-reviewed against the current head; superseding the stale requested-changes state.

@christian-byrne
christian-byrne merged commit d4413a0 into main Sep 27, 2026
73 checks passed
@christian-byrne
christian-byrne deleted the docs/qa-followup-tdd branch September 27, 2026 04:54
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

risk:medium PR risk grade (advisory shadow check; grader-owned)

Projects

None yet

Development

Successfully merging this pull request may close these issues.

5 participants