All supported GHES versions (3.16 through 3.20) support the
checkRunCountsByState and statusContextCountsByState fields on
StatusCheckRollupContextConnection. The slow path that iterated
individual CheckContext nodes in ChecksStatus() is dead code.
This commit:
- Removes the slow path from ChecksStatus(), keeping only the
aggregated counts-by-state path
- Removes parseCheckStatusFromCheckConclusionState (no callers remain)
- Removes CheckRunAndStatusContextCounts from PullRequestFeatures
and its introspection detection
- Consolidates the two feature detection introspection queries into
one (PullRequest + WorkflowRun fits within the platform limit of
two __type expressions)
- Removes the errgroup dependency from feature detection
- Always uses statusCheckRollupWithCountByState in pr status queries
- Updates pr view fixtures to include counts-by-state fields
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
When a GitHub Actions workflow uses concurrency with cancel-in-progress,
cancelled runs were counted as failures in `gh pr status` and
`gh pr view`, even when a newer run for the same check name succeeded.
The GitHub web UI and `gh pr checks` both handle this correctly.
Three changes fix this:
1. Add a `cancelled` check status category. Cancelled runs are now
excluded from all summary counts (passing/failing/pending) and
subtracted from the total, matching the web UI behavior.
2. Move `eliminateDuplicates` from pkg/cmd/pr/checks to
`api.EliminateDuplicateChecks` (exported). The function operates
entirely on `api.CheckContext` and is now shared by both `pr checks`
and `ChecksStatus()` (used by `pr status` and `pr view`).
3. Apply deduplication in the `ChecksStatus()` slow path, keeping only
the most recent run per check name — consistent with `pr checks`.
Fixes#12895
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Replace copilotDisplayName with actorDisplayName(typeName, login, name)
which handles all actor types: known bots get friendly names (e.g.
Copilot → 'Copilot (AI)'), regular bots return login, users with
names return 'login (Name)', others return login.
All DisplayName() methods on Author, CommentAuthor, GitHubUser,
AssignableUser, AssignableBot, RequestedReviewer, and ReviewerBot
now delegate to actorDisplayName with their available fields.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Add CopilotDisplayName helper that translates known Copilot bot
logins (copilot-pull-request-reviewer, copilot-swe-agent) to the
friendly 'Copilot (AI)' display name. Applied to:
- PullRequestReview.AuthorLogin() — review comment author
- Comment.AuthorLogin() — PR/issue comment author
- parseReviewers() in pr view — reviewer list display
This ensures gh pr view shows 'Copilot (AI)' instead of the raw
'copilot-pull-request-reviewer' login for both the reviewer status
line and any review comments left by Copilot.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
- add usage examples to --help output
- use path.Match instead of filepath.Match for OS-consistent behavior
- rename patterns to excludePatterns for clarity
- simplify splitDiffSections using strings.Split
Add a new --exclude (-e) flag to gh pr diff that allows users to exclude
files matching glob patterns from the diff output. This is useful for
filtering out auto-generated files, vendor directories, or other noise
when reviewing pull requests.
Supports standard glob patterns and can be specified multiple times.
Patterns match against both the full path and basename.
Closes#8739
Use DefaultLogins instead of Default display names when calling
MultiSelectWithSearch for reviewers. The dedup logic in the
prompter compares keys (logins) against defaults, so passing
display names like 'mxie (Melissa Xie)' prevented deduplication
against search result keys like 'mxie'.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Wire CopilotReviewerReplacer into NewIssueState so that
`gh pr create --reviewer @copilot` correctly resolves to the
copilot-pull-request-reviewer bot login, matching the behavior
already implemented in gh pr edit.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Mark the GHES ID-resolution branch in AddMetadataToIssueParams
for cleanup once GHES supports requestReviewsByLogin.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Team reviewers must be provided as fully qualified org/teamname.
Remove the /slug shorthand that auto-prefixed the repo owner,
as this format is not supported.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Mark the legacy static MultiSelect reviewer path for cleanup once
GHES supports requestReviewsByLogin and search-based selection can
be used universally.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Gate search-based reviewer selection on both state.ActorReviewers
and the search function being available, consistent with the
ActorAssignees pattern used for assignees.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Replace `1;38` with `1;37` (bold white) in the delimiter/header
color constant. SGR parameter 38 is the extended foreground color
prefix and requires sub-parameters (e.g. `38;5;n` or `38;2;r;g;b`),
so using it bare produces an invalid escape sequence. Most terminals
silently ignore the malformed parameter, masking the bug.
Fixescli/cli#12683
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Add greppable TODO identifiers above all if-statements that reference
featuredetection struct fields, as required by the featuredetection
linter. This ensures every feature detection branch is tagged for
future cleanup when GHES gains support.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Update comment in FetchOptions to specify that the APIs used for both GHES and GitHub.com accept user logins and team slugs directly, clarifying why non-interactive flows with Add/Remove don't need to fetch reviewers/teams. Comment-only change; no functional modifications.
Add clarifying comment in pkg/cmd/pr/edit/edit.go to note that missing assignee/reviewer search functions trigger a downstream fallback to legacy fetching. In pkg/cmd/pr/shared/editable.go remove a redundant line and add a TODO urging migration of non-interactive assignee updates to use the new logins input with ReplaceActorsForAssignable to avoid unnecessary fetching. These are comment and doc changes to clarify intent and future improvements.
Remove a redundant struct-level comment and update the DefaultLogins field comment in pkg/cmd/pr/shared/editable.go to more accurately describe its purpose: used to disambiguate actors from display names rather than to compute add/remove sets.
Always send explicit lists for userLogins, botLogins, and teamSlugs
in RequestReviewsByLogin mutation, even when empty. Previously, empty
slices were omitted due to omitempty JSON behavior and len > 0 checks,
which prevented clearing all reviewers when using replace mode.
Empty slices are harmless no-ops in union mode, so we now send them
unconditionally for simpler logic.
Add test to verify the mutation receives empty slices when all
reviewers are removed.
Enables Copilot to be requested as a pull request reviewer, supporting both interactive and non-interactive flows. Introduces new reviewer search functionality, updates reviewer partitioning to distinguish users, bots, and teams, and adds a GraphQL mutation for reviewer management on github.com. Updates help text, tests, and internal APIs to support Copilot reviewer login and display names, while maintaining compatibility with GitHub Enterprise Server.
Updated SuggestedAssignableActors to return the total count of available assignees in the repository. Modified assigneeSearchFunc to use this count to calculate and display the number of additional assignees beyond the current results.
Introduces a test case to verify that the interactive edit flow on GitHub Enterprise Server uses the legacy assignee selection without search, ensuring correct behavior when editing pull request assignees.
Added detailed comments to the assigneeSearchFunc explaining its purpose and the importance of updating assignable actors metadata for later ID resolution when mutating assignees with the GraphQL API.
Updated test mocks and logic to consistently use lowercase 'monalisa' for login names and display names for user assignees. Improved handling of dynamic assignee fetching in interactive flows by relying on searchFunc and metadata population, and clarified logic in FetchOptions to fetch assignees only when necessary. These changes ensure more accurate simulation of interactive assignment and better test coverage for actor assignee features.
Refactored the MultiSelectWithSearch function and related interfaces to use a MultiSelectSearchResult struct instead of multiple return values. This change improves clarity and extensibility of the search function signature, and updates all usages, mocks, and tests accordingly.
The assigneeSearchFunc now receives the editable struct to update its Metadata.AssignableActors field with suggested assignable actors. This change ensures that the editable struct has the necessary actor metadata for later PR updates.
Introduces SuggestedAssignableActors API query and wires up a dynamic assignee search function in the PR edit command. Updates Editable and EditPrompter interfaces to support search-based multi-select for assignees, improving the user experience when assigning users to pull requests.
