Repository navigation
fix(gitlab): merge request details report their head commit - #18608
Merged
Yash-Singh1 merged 1 commit intoOct 12, 2026
Merged
Conversation
A review guide needs the pull request's head commit from its detail. The GitLab detail decoder dropped the merge request's `sha`, so generating a guide for any GitLab merge request stopped with "This host did not report the pull request's head commit." The detail now carries `headSha` from GitLab's `sha`, and leaves it out where GitLab sends none. Refs pingdotgg#18592 Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Contributor
ApprovabilityVerdict: Approved at Macroscope's review found this PR approvable — This is a narrowly scoped GitLab decoder bug fix that propagates the existing merge request head SHA into already-implemented review-guide and watch behavior, with regression tests for missing and partial API responses. It does not change defaults, schemas, deployment configuration, or static-analysis settings. Notes:
You can add or adjust custom eligibility rules. Learn more. |
macroscopeapp
Bot
dismissed
their stale review
October 12, 2026 10:12
Dismissing prior approval to re-evaluate 328e80c
aorwall
added a commit
to aorwall/t3code
that referenced
this pull request
Oct 13, 2026
Merges upstream `pingdotgg/t3code` `5070a83a32` (48 commits, base `e688034b5f`) into the fork. It does not stack on another PR. ## Merge - 240 of 240 upstream-range files landed. Fork delta: 775 files. - One conflict: `apps/web/src/routeTree.gen.ts`, regenerated for upstream's new `/_chat/new` route. The other 18 files both sides touched auto-merged: `duplicate-adds` and `resolution-check` are clean, and every delta guard holds. - `pnpm-lock.yaml` was re-derived by install, and `verify --only lockfile` passes. - Sweep: `promptLinks/HostedPromptLinkLanding.tsx` renders only in the `hosted-static` auth state, which Moatless never reaches. Taken as upstream's. - Unsupported methods: ADD 0, DROP 0. ## Verification `verify.mjs` passes duplicate-adds, tripwires, resolution-check, unsupported-methods, lockfile, fmt, lint, typecheck and build. `test` fails only on files that fail alone and failed the same way on the last three merges, because of the sandbox: shared `AgentScope` (`oom_score_adj` is 935 here), server `ServerBrowserPage` (no Playwright browser), `AcpAdapterV2`, `OrchestratorReplayFixtures`, and provider-opencode `OpenCodeServerLedger`. ## Usable as-is - Prompt links (pingdotgg#18454): `/new?prompt=…`, `ThreadLinkPrompt` on thread routes, and copy-prompt-link from messages and the command palette. Choosing a project and placing the prompt work against Moatless. - Terminal colors (pingdotgg#17168), context meter rounding (pingdotgg#17539), stable button text (pingdotgg#18866), setup wizards follow themes (pingdotgg#18725), Git actions follow the selected thread (pingdotgg#18862), dropped threads attach to pending answers (pingdotgg#18789), the compact merge-button second click (pingdotgg#18878), a numeric `q` on pull requests (pingdotgg#18666), the home-page bounce fix (pingdotgg#18672), running terminals shown in the sidebar (pingdotgg#18740), and computer-use titles (pingdotgg#18865). - Mobile and desktop fixes (theme contrast, settings rows, the PR action bar, image caching, the iOS 27.2 black screen, Hyprland, updater diagnostics). - "Update everything" sidebar pill (pingdotgg#18053): inert against Moatless, which advertises no `serverSelfUpdate` or `versionAdvisory`. ## Unsupported in Moatless / needs implementation - Prompt-link "clone this repository" (`NewThreadLinkView.cloneRepository`): needs `projectClone.start` and the `projectCloneTracking` capability. It refuses with a message today. Added to gaps.md under *Project and repository management*. - Renaming built-in models (pingdotgg#18840): saves through `server.updateSettings` (custom model entry with a built-in slug). It stays behind `FEATURES.providerConfiguration`, under the existing settings gap. - Provider secret settings (pingdotgg#18505, `secretSetting` and `isSecretSetting` in contracts): stored on the host, so they fall under the same settings gap. ## Backend behavior to consider reproducing in Moatless Recorded in gaps.md under *Runtime fixes upstream made to its own server*: - A confirmed Claude subagent outcome settles that task and its child work (pingdotgg#18778, the new `subagent` target on `thread.background-work.settle`). - Claude thinking before a tool call stays a collapsed Thought (pingdotgg#18887). - Codex title and branch-name generation no longer starts the user's MCP servers (pingdotgg#17058). - The event store resyncs a thread holding an event type from a newer build (pingdotgg#15937, `isKnownOrchestrationV2EventType`). - A migration id clash stops startup with a clear error (pingdotgg#15925). - Cached branch PR lookups start no git process (pingdotgg#17063). - PR head commits are reported for Forgejo, GitLab and Bitbucket (pingdotgg#18463, pingdotgg#18608, pingdotgg#18402), and PR bases resolve to branch names (pingdotgg#18754). 🤖 Generated with [Claude Code](https://claude.com/claude-code) --- Moatless task: https://moatless.soaplabstest.com/tasks/dcd13b46-6495-4b1d-af7c-03da83644d11
adampeterhiggins
added a commit
to adampeterhiggins/t3code
that referenced
this pull request
Oct 13, 2026
* fix(gitlab): merge request details report their head commit (pingdotgg#18608) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> * feat(server): rename built-in models from settings (pingdotgg#18840) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> * fix(web): Git actions follow the selected thread (pingdotgg#18862) * fix(desktop): preserve updater failure causes in diagnostics (pingdotgg#18864) * fix(web): attach dropped threads to pending answers (pingdotgg#18789) * fix(web): keep button text stable during press scaling (pingdotgg#18866) * fix(bitbucket): preserve PR head commits for review guides (pingdotgg#18402) * fix(server): return null for unresolved optional commands (pingdotgg#18871) * fix: show computer-use action titles and align app icons (pingdotgg#18865) * fix(server): apply confirmed Claude subagent outcomes (pingdotgg#18778) * fix(mobile): Browser page no longer logs a duplicate key warning (pingdotgg#18776) * fix(web): a numeric q in the pull requests URL no longer crashes the page (pingdotgg#18666) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> * fix(client): T3 Connect clients reconnect once a new tunnel is ready (pingdotgg#16554) Co-authored-by: PR Batch Tester <agent@local.test> Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * fix(web): leaving the home page while the app loads no longer bounces you to a new draft (pingdotgg#18672) Co-authored-by: PR Batch Tester <agent@local.test> Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> * fix(mobile): attachment images no longer re-download on every reconnect (pingdotgg#18824) Co-authored-by: PR Batch Tester <agent@local.test> Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> * fix(server): a migration id clash stops startup with a clear error instead of crash-looping (pingdotgg#15925) Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Co-authored-by: PR Batch Tester <agent@local.test> * fix(server): a thread with an event type from a newer build can still resync after a downgrade (pingdotgg#15937) Co-authored-by: Claude Sonnet 5 <noreply@anthropic.com> Co-authored-by: PR Batch Tester <agent@local.test> * fix(server): Codex titles and branch names no longer start the user's MCP servers (pingdotgg#17058) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> Co-authored-by: Yash Singh <saiansh2525@gmail.com> * fix(server): Claude thinking before a tool call stays a collapsed Thought (pingdotgg#18887) Co-authored-by: maria-rcks <254055478+maria-rcks@users.noreply.github.com> * perf(server): cached branch PR lookups no longer start git processes (pingdotgg#17063) * fix(web): compact merge buttons show when they need a second click (pingdotgg#18878) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> * fix(mobile): pull requests load from another environment when theirs is offline (pingdotgg#18698) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> * perf(web): collect active tool calls without array shifts (pingdotgg#18853) Co-authored-by: SAPHID <saphid@users.noreply.github.com> Co-authored-by: GPT-6.1 Sol <noreply@openai.com> * fix(web): center the pace icon on the limit card's "left" label (pingdotgg#18580) * fix(web): update everything includes npm-installed servers again (pingdotgg#18709) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> * fix(server): review guides find Bitbucket heads reported by short hash (pingdotgg#18900) Co-authored-by: Pawel Mikolajczuk <pmikolajczuk@atlassian.com> Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> * fix(web): scheduled task model picker no longer shifts the dialog (pingdotgg#18940) * fix(ci): CodeRabbit reviews no longer request changes (pingdotgg#18954) --------- Co-authored-by: Scott Norteman <snorteman@gmail.com> Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> Co-authored-by: Theo Browne <me@t3.gg> Co-authored-by: Yash Singh <saiansh2525@gmail.com> Co-authored-by: Paweł Mikołajczuk <pawel@mikolajczuk.in> Co-authored-by: Tristan Knight <admin@snappeh.com> Co-authored-by: shivam <91240327+shivamhwp@users.noreply.github.com> Co-authored-by: PR Batch Tester <agent@local.test> Co-authored-by: Bryce Bjork <brycedbjork@gmail.com> Co-authored-by: maria <maria@kuuro.net> Co-authored-by: maria-rcks <254055478+maria-rcks@users.noreply.github.com> Co-authored-by: Dara Adedeji <76637177+SunkenInTime@users.noreply.github.com> Co-authored-by: Antony <97451137+TonybynMp4@users.noreply.github.com> Co-authored-by: Julius Marminge <julius0216@outlook.com> Co-authored-by: Alex Southwell <saphid@gmail.com> Co-authored-by: SAPHID <saphid@users.noreply.github.com> Co-authored-by: GPT-6.1 Sol <noreply@openai.com> Co-authored-by: Akash Moradiya <64416825+akash3444@users.noreply.github.com> Co-authored-by: Pawel Mikolajczuk <pmikolajczuk@atlassian.com> Co-authored-by: oliver <97427849+flamboh@users.noreply.github.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Problem
Generating a review guide for any GitLab merge request fails with "This host did not report the pull request's head commit." The guide service reads the head commit from the pull request detail, and the GitLab detail decoder (
gitLabMergeRequestJson.ts) dropped the merge request'ssha.Change
The raw merge request schema now reads GitLab's top-level
sha, the source branch's head commit, andtoDetailpasses it on asheadSha. When GitLab sends none,headShais left out.GitLabPullRequestProvider.getChangeRequestalready spreads the decoded detail, so the provider needs no change. This is the GitLab counterpart of the Forgejo fix in #18463 and the Bitbucket fix in #18402.The fix reads
sharather thandiff_refs.head_shabecause GitLab fills indiff_refsonly after it computes the diff, so it can be null on a new merge request. The list mapper is unchanged.The pull request watch also reads the detail's
headSha, so it can now tell when a watched GitLab merge request's head moves, as it already does for GitHub and Forgejo.Scope and approval
Fixes the bug confirmed in the maintainer triage: #18592 (comment)
Closes #18592
Verification
Agent-run on Linux. No person tested this.
Live run in headless Chromium (1280×800 viewport, shots cropped) against a Linux dev server signed in to gitlab.com with a read-only token, on two open gitlab-org/cli merge requests. Nothing was written to GitLab.
On main, Write guide shows "Could not start a review guide / This host did not report the pull request's head commit." and no run starts. On this branch the guide is written, and the head commit it stores matches GitLab's
shafor each merge request (772a4f2andafce4da). !4089's source branch is in a fork, and the guide still checked out its head. Its shot is from Rewrite guide, because a guide already existed there. The guides shown were written by Claude Sonnet 5.5.Main, !4064:
18592-main-4064-toast.mp4
This branch, !4064:
18592-fix-4064-guide.mp4
vp test runinpackages/source-control-gitlab: 136 passed. The new case ("reads the head commit, and says nothing where GitLab omitted it") fails on main, whereheadShaisundefined.vp lintandvp fmt --checkon the two changed files, and thesource-control-gitlaband server typechecks, pass.Model: Claude Opus 5.5 (implementation and review), GPT-6-Astra (review). Harness: Claude Code, Codex.