[mirror] microsoft/vscode#299072 Fix compile error in mcpGatewayToolBrokerChannel test#468
Open
austindyoung wants to merge 693 commits intomainfrom
Open
[mirror] microsoft/vscode#299072 Fix compile error in mcpGatewayToolBrokerChannel test#468austindyoung wants to merge 693 commits intomainfrom
austindyoung wants to merge 693 commits intomainfrom
Conversation
…of domains. (microsoft#298372) * fixing when trusted domains are url's instead of just domains * changes * changes
* set remote repo uri * fix setting repo * fix setting repo uri
As no editor can open non-modal, this is not helping much.
…8409) * sessions - show command in toast and show it more often * Update src/vs/workbench/contrib/chat/browser/chatWindowNotifier.ts Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com> --------- Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
* sessions - tweaks to chat input to make it larger * Update src/vs/sessions/browser/media/style.css Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com> --------- Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
…opened (microsoft#297783) * Git - file-system provider should open the repository if not already opened * Git - only open the repository in the empty/session window * Add logging
* Clean up model picker input styles: remove background, add filter spacing * Use CSS :has() instead of imperative class for filter container
* Chat input refinements: increase height and update placeholder text * Extract shared constants for input editor line height and padding
* Report Event.Buffer leaks when running from source * Address feedback
…leted and changed (microsoft#298272) * updating the edit context on content change not on lines inserted, deleted and changed * saving the previous edit context text
Inserting `text-spacing-trim: space-all;` for `.monaco-editor`
…vers (microsoft#298484) Add a new setting `editor.hover.showLongLineWarning` (default: true) that controls whether the "Tokenization is skipped..." and "Rendering paused..." hover messages are shown on long lines. When these hovers are shown, they now include a "Don'\''t Show Again" link that sets the setting to false, letting users dismiss them permanently while keeping all other hover functionality intact. Fixes microsoft#172713
* Plumb 'isBuiltin' for chat modes into request microsoft/vscode-internalbacklog#6884 * Fix test
…oft#298497) Copilot-authored message: Fix inline reference at block start rendering on its own line When an inlineReference is the first item in a chat response (no preceding markdown), it creates a standalone markdownContent with default MarkdownString properties. The subsequent markdown from the model has different properties (e.g. isTrusted: true), causing canMergeMarkdownStrings to return false. This leaves them as separate content parts, rendering the inline reference link on its own line. Fix by detecting when the previous item consists solely of synthesized content-ref links (via isContentRefOnly check) and merging them by prepending the reference text while adopting the incoming markdown's properties. Fixes microsoft#278191
* editor: add 'foldedLine' unit to cursorMove command Add a new 'foldedLine' movement unit to the `cursorMove` command that moves by model lines while treating each folded region as a single step. When moving down/up by 'wrappedLine' the cursor skips folds naturally because it operates in view space. When moving by 'line' it uses model coordinates and can land inside a fold, causing VS Code to auto-unfold it. The new 'foldedLine' unit moves in model space but queries `viewModel.getHiddenAreas()` to detect folds and jump to the first visible line past each one, so folds are skipped without being opened. This is the semantics needed by vim's j/k motions (VSCodeVim/Vim#1004): each fold counts as exactly one step, matching how real vim treats folds. Fixes: VSCodeVim/Vim#1004 * editor: simplify foldedLine movement using fold-walk algorithm Replace the step-by-step simulation (O(count × folds)) with a single pass over sorted hidden areas (O(folds in path)). Compute a naive target, then extend it for each fold encountered, stopping before any fold that reaches the document boundary. Also extracts _targetFoldedDown/_targetFoldedUp helpers to eliminate the duplicated loop structure between the two directions. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * editor: fix foldedLine count movement at fold boundaries --------- Co-authored-by: Jai <jai@jai.one> Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com> Co-authored-by: Alexandru Dima <alexdima@microsoft.com>
Better handle event subscriptions (microsoft#293200)
…98508) * fix microsoft#288515: wrong current index in find widget if matches > 1000 When there are more than 1000 matches, decorations use _FIND_MATCH_NO_OVERVIEW_DECORATION instead of _FIND_MATCH_DECORATION. getCurrentMatchesPosition did not check for this decoration type, causing it to return 0 and fall through to a fallback with an off-by-one error. * Address review feedback * Fix unused error
…8512) * Fix potential listener leak in document semantic tokens Hoist provider onDidChange and registry onDidChange subscriptions from each ModelSemanticColoring instance into the singleton DocumentSemanticTokensFeature. Previously, every ModelSemanticColoring subscribed individually to both the global LanguageFeatureRegistry.onDidChange and each provider's onDidChange event, resulting in O(N*M) listeners (N models × M providers). In scenarios like chat editing where many models are created rapidly, these listeners accumulated and triggered leak detection. Now the singleton subscribes once to the registry change and once per provider (via allNoModel()), then fans out notifications to watchers. Each watcher checks provider relevance via _provider.all(model).includes() before acting on the event. Also replaces manual IDisposable[] management with a DisposableStore for proper lifecycle tracking. * Review feedback
* fix non-edit codeblocks in thinking * Address some comments
Per review feedback: Outdated servers should use the same per-server startup grace period as Unknown servers. A prior fast startup does not guarantee a fast restart, so both states now race startup against the 5s timeout before returning results.
docs: fix duplicated wording in proposed API comment\n\nSigned-off-by: Rohan Santhosh <rohan@example.com>
…-2026-theme-syntax-colors Update syntax colors for 2026 theme
…-walrus Fix cut off corner border
* Polish question carousel: keyboard nav, badge styling, focus outlines * chat: prioritize selected single-select answer over freeform draft * Enhance question carousel with tab navigation and review functionality * Polish question carousel: stack title/desc, plain numbers, multiline Q:/A: summary, Cmd+Enter submit * Polish question carousel titles and execute controls * Address PR feedback: guard checkmark on actual answers, restore queue/steer, deduplicate format helper, fix JSDoc * Fix carousel tests: update selectors for tab-bar UI * Fix remaining carousel tests from main merge * Add ARIA tabpanel pattern and clear tab indicators on dispose * Adjust tab bar padding-left to 4px for multi-question carousels * Fix CI: non-null assertion and remove unused constants * fix: show steer/queue submenu during question carousel when text is typed - Cancel button now hides when input has text during question carousel (matches behavior during regular in-progress requests) - Queue/steer submenu now appears during question carousel and tool confirmation states, since requestInProgress is false in those states - Removed unused requestInProgressWithoutInput and pendingToolCall vars
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
Add very basic documentation of built-in extension structure
Adds support for managing tool confirmation preferences directly from the tool picker. This allows users to approve tools and external paths at the workspace level, reducing friction when using tools that require confirmation. - Adds 'Manage Approval' button to tools that support confirmation - Integrates ILanguageModelToolsConfirmationService with tool picker - Adds workspace-level allowlist persistence for external paths - Extends ActionableButton type to support keepOpen behavior - Implements workspace folder selection and allowlist management - Adds ObservableMemento for persistent storage of approved paths Fixes microsoft/vscode-internalbacklog#6805
…icrosoft#299048) Adds max-height: 90vh to the dialog box to prevent it from exceeding the viewport height, matching the existing max-width: 90vw constraint. The message content area now scrolls when it exceeds available space: - Added align-self: stretch to .dialog-message-container in horizontal layout so it fills the row height and triggers overflow-y scrolling - Added min-height: 0 to .dialog-message-container in vertical layout to allow flex shrinking and overflow-y scrolling - Changed .dialog-message-container overflow from 'hidden' to 'overflow-y: auto; overflow-x: hidden' to enable vertical scrolling Buttons, toolbar, and footer remain visible. Only message content scrolls. Fixes microsoft#296528 (Commit message generated by Copilot)
* chat: support relative paths in plugin locations - Changes chat.pluginLocations to support home-relative (~/) and workspace-relative (./) paths instead of absolute paths - Adds path normalization and validation in agentPluginServiceImpl - Updates constants and contribution for plugin path handling - Ensures settings are shareable across machines by avoiding absolute paths Fixes microsoft#297365 (Commit message generated by Copilot) * comment
microsoft#299030) sessions: add "github.copilot.chat.githubMcpServer.enabled" to config override
…soft#299035) Per the MCP spec, the 'properties' field in tool inputSchema is optional, with 'type' being the only required field. However, JSON Schema Draft 7 validation requires 'properties' for object types, causing spurious warnings like 'Tool foo failed validation: schema must have a properties object'. Fix by normalizing the inputSchema to include an empty properties object when not present. Fixes microsoft#251723
…ft#299052) * chat: add manage action gear icon to installed agent plugins Adds a ManagePluginAction that displays a gear icon in the plugin list for installed plugins. Clicking the gear icon shows a context menu with management options (enable/disable, open folder, open README, uninstall), making it consistent with how extensions and MCP servers display their management UI. - Adds ManagePluginAction class that provides dropdown menu with management actions for installed plugins - Adds ManagePluginActionViewItem to render the action as a clickable gear icon with context menu - Updates AgentPluginRenderer to include the gear icon in the action bar for installed plugins - Context menu displays enable/disable, open folder, open README, and uninstall options Fixes microsoft#298461 (Commit message generated by Copilot) * chat: clean up manage plugin action to follow MCP pattern - Extract shared getInstalledPluginContextMenuActionGroups() to deduplicate action construction between gear menu and context menu - Restructure ManagePluginAction with createActionViewItem() pattern matching MCP/extensions DropDownAction approach - Add onHide disposal for context menu actions - Remove redundant getDomNodePagePosition import
…icrosoft#299061) * chat: combine multiple pending steering messages into single request When Copilot is running a tool, sending multiple steering messages only sends the first one. The chatbot waits indefinitely for the remaining messages. This fix combines all consecutive pending steering messages into a single request by: - Adding dequeueAllSteeringRequests() to ChatModel to dequeue all consecutive steering messages at the front of the queue - Refactoring processNextPendingRequest() to handle both steering and queued requests through a unified flow that: - Combines multiple steering message texts with \n\n separator - Merges attachments from all steering messages - Re-parses the combined text - Sends as a single request to the agent Fixes microsoft#298324 (Commit message generated by Copilot) * comments
* McpGateway extensive logging * update * update
…race invalidation
…teway-compile-error
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
20 participants
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.
Automated mirror PR for DriftFence counterfactual telemetry.
Source PR: microsoft#299072
Source URL: microsoft#299072
Mirror branch: driftfence/mirror/pr-299072/latest
This PR is maintained by automation for telemetry and review links.