Some tests need real OS threading and timers rather than the
deterministic, single-threaded behavior of `TestDispatcher`. The
existing multithreaded dispatcher provides those production-like
scheduling semantics, but it is named and gated as a benchmark-only
facility, which unnecessarily pulls in Criterion and histogram
dependencies.
This renames `BenchDispatcher` to `ThreadedDispatcher` and exposes it
through GPUI's `test-support` feature. Benchmark contexts continue using
the same dispatcher, while Criterion and benchmark reporting remain
isolated behind the `bench` feature.
### Testing
- `cargo check -p gpui --features test-support`
- `cargo check -p gpui --features bench`
- `cargo nextest run -p gpui 'threaded_dispatcher::tests'`
- `cargo fmt --all --check`
Release Notes:
- N/A
# Objective
Add explicit conversation compaction for OpenAI Responses API models,
available through both direct OpenAI connections and Zed Cloud models.
Keep finalized replacement context provider-neutral and separate from
the lifecycle events emitted while automatic compaction is in progress.
## Solution
- Add opt-in explicit compaction support to the `LanguageModel` trait.
- Separate streamed `CompactionUpdate` lifecycle events from finalized
`CompactedContext`.
- Represent provider-native compacted context as an opaque, versioned
state owned by the originating provider.
- Construct compact requests through the same OpenAI request conversion
used for completions, then narrow the request to fields supported by
`/responses/compact`.
- Preserve and validate the complete canonical replacement window
returned by OpenAI.
- Replay the replacement window before messages sent after compaction,
without resending the superseded transcript.
- Support explicit compaction through both the direct OpenAI provider
and `CloudLanguageModel`.
- Update automatic OpenAI and Anthropic compaction to use the same
finalized context representation.
- Reject empty, malformed, incorrectly versioned, and incorrectly owned
provider state.
The corresponding Cloud change adds the `/completions/compact` proxy
route: https://github.com/zed-industries/cloud/pull/3099.
The downstream Delta UI integration is
https://github.com/zed-industries/delta/pull/1739.
## Testing
- `cargo nextest run -p anthropic -p open_ai -p language_models_cloud`
- `cargo check -p language_models --tests`
- `cargo fmt --all --check`
- `git diff --check`
The focused tests cover request construction, canonical-window
preservation and replay, direct and Cloud transports, automatic
compaction lifecycle events, malformed provider state, provider
ownership, format compatibility, and HTTP failure handling.
## Self-Review Checklist:
- [x] I've reviewed my own diff for quality, security, and reliability
- [x] Unsafe blocks (if any) have justifying comments
- [x] The content adheres to Zed's UI standards
([UX/UI](https://github.com/zed-industries/zed/blob/main/CONTRIBUTING.md#uiux-checklist)
and
[icon](https://github.com/zed-industries/zed/blob/main/crates/icons/README.md)
guidelines)
- [x] Tests cover the new/changed behavior
- [x] Performance impact has been considered and is acceptable
---
Release Notes:
- Added: System messages are now sent through the `instructions` field
in OpenAI Responses API requests.
Implements the previously-`unimplemented!()`
`TestPlatform::prompt_for_paths` so tests can drive the platform Open
dialog deterministically.
Adds `TestAppContext::simulate_path_prompt_response` and
`did_prompt_for_paths`, mirroring the existing `prompt_for_new_path`
test helpers (`simulate_new_path_selection`). The simulated response
validates that callers don't return multiple paths when
`PathPromptOptions::multiple` is false.
Release Notes:
- N/A
Adds `scheduler::spawn_dedicated_thread` (and inherent `spawn_dedicated`
methods on `PlatformScheduler` and `TestScheduler`) so single-threaded
actors that own `!Send` state can run on their own OS thread and freely
do blocking I/O without disturbing any other executor.
### Why
A single-threaded actor that needs to do blocking syscalls is currently
stuck: it can't run on the shared foreground executor (blocking would
stall every other foreground session), and it can't move to the
background pool because its state isn't `Send`. `spawn_dedicated` gives
each such actor its own thread and its own `LocalExecutor`, while still
participating in the same testable scheduler infrastructure as
everything else.
### Shape
- `pub fn spawn_dedicated_thread(session_id, scheduler, f) -> Task<_>`
in `scheduler`. Owns the OS thread, the per-session runnable channel,
and the `LocalExecutor` setup.
- Inherent `spawn_dedicated` on `PlatformScheduler` (allocates its own
`SessionId`, delegates to the free function).
- Inherent `spawn_dedicated` on `TestScheduler` (no real thread — runs
as a fresh local session driven by the test scheduler's run loop, so
determinism under `many` is preserved).
- Renames `Scheduler::schedule_foreground` → `schedule_local` and
`scheduler::ForegroundExecutor` → `scheduler::LocalExecutor` to reflect
that these are session-pinned queues rather than "the main thread" (a
dedicated session runs on its own thread). GPUI's wrapper
`gpui::ForegroundExecutor` and the `foreground_executor` field/method
names are unchanged to keep blast radius small.
- `LocalExecutor::new` now takes an explicit dispatch closure, so the
routing decision (default session, dedicated thread, or something else)
lives at the construction site.
### Tests
- `TestScheduler` side: round-trip, `!Send` future, `Send` closure
capturing shared state, inner `executor.spawn`, determinism under `many`
seeds, drop-cancels-future, detached child runs after root completes.
- `PlatformScheduler` side: real separate thread (blocking syscalls
don't stall the test), `!Send` future output, drop-cancels-future,
thread tears down after work completes, detached child outlives root.
cc @as-cii
Release Notes:
- N/A
---------
Co-authored-by: Antonio Scandurra <me@as-cii.com>
Co-authored-by: Conrad Irwin <conrad.irwin@gmail.com>
Closes#55619
### Summary
- Route `buffer_search::UseSelectionForFind` through
`BufferSearchBar::deploy` instead of updating the query editor directly.
- Add an explicit seed-query override to `deploy`, so the Cmd-E action
can force `SeedQuerySetting::Always` while regular deploy callers
continue to pass `None` and respect the user’s
`seed_search_query_from_cursor` setting.
- By going through `deploy`, Cmd-E now also runs the search path that
keeps buffer-search navigation state in sync:
- shows/initializes the search bar for the active searchable item
- applies the seeded query via `search_suggested`
- calls `search`, which updates the query editor, search options, active
search query, search history, and macOS find pasteboard
- refreshes `searchable_items_with_matches` and `active_match_index`
- activates the current match after the search completes
- This ensures the subsequent Cmd-G action has the expected active
query, match list, search token, and active match index to select the
next result.
- Add a macOS-only end-to-end regression test using the default macOS
keymap with `simulate_keystrokes("cmd-e")` and
`simulate_keystrokes("cmd-g")`.
### Validation
- `cargo test -p search test_cmd_e_then_cmd_g_uses_selection_for_find`
- `cargo fmt --check --package search --package zed_actions`
- `./script/check-keymaps`
- `cargo check -p search`
- `cargo check -p workspace`
- `cargo check -p vim`
Release Notes:
- Fixed macOS Cmd-E/Cmd-G find behavior so Cmd-E seeds find from the
cursor or selection and Cmd-G advances through the newly seeded matches.
This renames the Archive view to Thread History in all user-facing
surfaces. The concept of archiving a thread (the verb/state) remains
unchanged — only the view that lists all threads is renamed, since it
shows both active and archived entries.
## Changes
- Rename `ViewAllThreads` action → `ToggleThreadHistory`
- Context-sensitive tooltip: "Show Thread History" / "Hide Thread
History"
- Update action doc comment to reference "history" instead of "archive
view"
- Telemetry event: `Thread History Viewed`
- `SerializedSidebarView::Archive` → `History` (with `#[serde(alias =
"Archive")]` for backward compat)
- Add a Lucide-based clock icon adapted to 16×16 / 1.2px stroke
- Switch the history toggle button to use the new clock icon
- Update all three platform keymaps
cc @danilo-leal
Release Notes:
- Renamed the threads Archive view to Thread History and updated its
icon to a clock.
---------
Co-authored-by: Danilo Leal <daniloleal09@gmail.com>
## Summary
Adds three tests that exercise session disassociation ("Session not
found" after reopening a thread). The third test **fails**, reproducing
the bug.
### Root cause
A retained `ConversationView` with a `thread_error` receives
`AgentServersUpdated`, which triggers `reset()` → `close_all_sessions` →
state transitions to `Loading`. When the user then reopens the same
session via `open_thread`, `load_agent_thread`'s `has_session` check
returns `false` (because the retained view is in `Loading` state, not
`Connected`), so a **second** `ConversationView` is created for the same
session. Both views' async load tasks complete and both transition to
`Connected` with the same session ID. When the retained view is later
cleaned up, its `on_release` → `close_all_sessions` removes the session
from the connection, leaving the active view stranded with a dead
session.
### Tests
| Test | Status | What it verifies |
|------|--------|-----------------|
| `test_open_thread_on_visible_session_does_not_disassociate_it` | ✅
Pass | `load_agent_thread` short-circuits when session is already active
|
| `test_retained_thread_cleanup_does_not_disassociate_reopened_session`
| ✅ Pass | Eviction + reopen cycle works (close-then-reload) |
| `test_retained_thread_reset_race_disassociates_session` | ❌ **Fail** |
Reproduces the race described above |
cc @bennetbo @ConradIrwin
Release Notes:
- N/A
---------
Co-authored-by: Ben Brandt <benjamin.j.brandt@gmail.com>
Co-authored-by: Bennet Bo Fenner <bennetbo@gmx.de>
Remove the `DEFAULT_THREADS_SHOWN` limit and the View More / Collapse
row from the sidebar thread list. When a project group is expanded, all
of its threads are now shown. Fold/unfold of project group headers is
preserved.
### Changes
- Remove `ListEntry::ViewMore` variant and `render_view_more`
- Remove `DEFAULT_THREADS_SHOWN`, `group_extra_batches`,
`set_group_visible_thread_count`, `expand/collapse_thread_group`
- Remove `visible_thread_count` from `ProjectGroupState`,
`SerializedProjectGroupState`, `ProjectGroup`, and persistence
- Remove `ShowMoreThreads` / `ShowFewerThreads` actions and vim
keybindings
cc @danilo-leal
Release Notes:
- N/A
Remove the ability to add and remove individual folders from project
groups in the sidebar.
## Changes
**Sidebar project group menu:**
- When there are 2+ local project groups, the ellipsis menu shows "Open
Project in New Window" and "Remove Project" (no separators, no folder
listing)
- When there's only one group or it's remote, the ellipsis menu is
replaced with a close X button that removes the project directly
**Project picker labels:**
- "Open Local Project" → "Open Local Folders"
- "Open Remote Project" → "Open Remote Folder"
- "Add Local Project" → "Add Local Folders"
- "Add Remote Project" → "Add Remote Folder"
**Dead code removal:**
- `MultiWorkspace::remove_folder_from_project_group`
- `MultiWorkspace::prompt_to_add_folders_to_project_group`
- `MultiWorkspace::add_folders_to_project_group`
- `MultiWorkspace::update_project_group_key`
- `MultiWorkspaceEvent::ProjectGroupKeyUpdated`
- `Sidebar::move_threads_for_key_change`
- `ThreadMetadataStore::change_worktree_paths_by_main`
- Associated tests for removed functionality
Per-workspace thread regrouping (when a workspace's worktree paths
change) continues to work via the existing `WorktreePathsChanged` →
`move_thread_paths` flow. The existing
`test_worktree_add_only_regroups_threads_for_changed_workspace` test
(renamed and strengthened) verifies that only the changed workspace's
threads are regrouped, while sibling worktree threads remain in their
original group.
Release Notes:
- Improved sidebar project group menu to show a simple close button for
single-project windows and a streamlined two-item menu for multi-project
windows.
When reopening a thread that is already visible in the agent panel,
`open_thread` could recreate the session instead of reusing the existing
visible thread. That left the UI holding a thread whose backend session
could be closed out from under it.
This change routes `AgentPanel::open_thread` through
`load_agent_thread`, which already handles the right behavior for
existing sessions:
- reuse the active thread when it already matches
- promote a retained thread when it matches
- only load when the session is not already present
The regression coverage is now a focused behavioral test at the
`AgentPanel::open_thread` boundary that verifies reopening an
already-visible session keeps the thread usable.
cc @bennetbo
Release Notes:
- Fixed reopening an already-visible agent thread so it reuses the
existing session instead of creating a broken duplicate.
This was leftover from trying to hold the invariant of one empty draft,
which was dropped in #53852.
Removes all automatic draft deletion:
- `retain_running_thread` discarded empty drafts when navigating away
- `remove_empty_draft` cleaned up all retained drafts when loading from
history
- `open_external_thread_with_server` removed empty drafts when opening
external threads
Release Notes:
- Fixed agent drafts being silently discarded when navigating between
threads
Previously, clicking a project header in the agent sidebar would
activate/switch to that worktree's workspace, resuming whatever thread
was most recently active. This felt jarring — activating a random thread
based on recency isn't that useful when the user can just click the
specific thread they want. There's also no clear semantic for clicking
the overall group header since each thread maps to a specific workspace
and there's no single workspace to activate for the group.
This changes the click target for folding from just the chevron to the
entire header row. The worktree-switching behavior is removed — clicking
anywhere on the row now toggles the fold, which is a more predictable
and useful interaction.
The hover-only buttons on the right side (ellipsis menu, collapse
threads, new thread) continue to work as before, with their own click
handlers that stop propagation.
Release Notes:
- Improved sidebar project headers to toggle fold on click instead of
switching worktrees.
Previously, the sidebar would automatically create an empty draft thread
for any project group that had no threads (via `reconcile_groups`). This
added complexity and felt janky — the user would see a phantom "Draft"
entry appear for every workspace they opened, even before they intended
to start a conversation.
Now, project groups can simply be empty. When a group has no threads:
- The fold indicator renders as closed and grayed out (disabled — no
click handler or tooltip)
- The plus button is always visible on the right side of the header (not
hidden behind hover)
- Keyboard fold toggling is disabled for the group
- Clicking the plus button creates the first thread and expands the
group
Draft threads are no longer auto-created anywhere:
- `AgentPanel::load` no longer creates a draft when there is no
previously active conversation
- Archiving the last thread in a group clears the panel instead of
creating a replacement draft
- `show_or_create_empty_draft` and `clear_active_thread` are removed
This removes a significant amount of supporting infrastructure that only
existed to maintain the "no group may be empty" invariant:
- `WorkspaceSidebarDelegate` trait and `AgentPanelSidebarDelegate`
struct
- `reconcile_groups` method and its 12 call sites
- `clear_empty_group_drafts` method and its 3 call sites
- `show_or_create_empty_draft` and `clear_active_thread` methods
- `reconciling` field on `Sidebar`
- `sidebar_delegate` field, setter, and getter on `Workspace`
Release Notes:
- Improved the sidebar to show empty project groups with a visible plus
button instead of auto-creating placeholder draft threads.
---------
Co-authored-by: Zed Zippy <234243425+zed-zippy[bot]@users.noreply.github.com>
Co-authored-by: Max Brunsfeld <maxbrunsfeld@gmail.com>
> **Foundation PRs:** #53732, #53733, #53734 — These three draft PRs
contain the base refactors (retained workspaces, agent panel overlay
split, ThreadId introduction) that this PR builds on.
## Goal
Remove `DraftId` and merge `draft_threads` + `background_threads` into a
single `retained_threads: HashMap<ThreadId, Entity<ConversationView>>`
in `AgentPanel`. A draft is just a thread that hasn't sent its first
message — no separate identity or storage needed.
## Changes
### agent_panel.rs
- Remove `DraftId` / `DraftIdCounter` / the `Global` impl
- Merge the two maps into `retained_threads`
- Add `thread_id: ThreadId` field to `BaseView::AgentThread`
- Rename methods: `create_draft` → `create_thread`, `activate_draft` →
`activate_retained_thread`, `remove_draft` → `remove_thread`, etc.
- Replace `clear_active_thread` with `show_or_create_empty_draft`
- Update `update_thread_work_dirs` to sync `ThreadMetadataStore` paths
when worktrees change
- Keep `load_agent_thread(...)` cleanup so activating a real thread
removes empty retained drafts
### sidebar.rs
- Remove `active_entry` derivation from `rebuild_contents` (was racing
with deferred effects)
- Add `sync_active_entry_from_panel` called from event handlers instead
- Simplify `ActiveEntry` — remove `ThreadActivation` struct, make
`session_id` optional
- Move `seen_thread_ids` to global scope (was per-group, causing
duplicate thread entries)
- Remove dead code: `clear_draft`, `render_draft_thread`
- Generalize `pending_remote_thread_activation` into
`pending_thread_activation` so all persisted-thread activations suppress
fallback draft reconciliation
- Set the activation guard before local persisted-thread activation
switches workspaces
- Make `reconcile_groups(...)` bail while a persisted-thread activation
is in flight
- On `ActiveViewChanged`, clear empty group drafts as soon as a pending
persisted-thread activation resolves
### thread_metadata_store.rs
- Make `ThreadMetadata.title` an `Option<SharedString>` with
`display_title()` fallback
- Add `update_worktree_paths` for batched path updates when project
worktrees change
### Other crates
- Update all `ThreadMetadata` construction sites and title display sites
across `agent_ui` and `sidebar`
## Fallback draft invariant
This PR now tightens the retained-thread invariant around fallback
drafts vs real thread activation/restoration:
- Fallback drafts are always empty
- User-created drafts worth preserving are non-empty
- While a persisted thread is being activated/restored/loaded, sidebar
reconciliation must not create an empty fallback draft for that target
group/workspace
- Once the real thread becomes active, empty fallback drafts in that
target group are removed
This is enforced by the sidebar-side activation guard plus existing
`AgentPanel` empty-draft cleanup after real-thread load.
## Tests
Added and/or kept focused sidebar regression coverage for:
-
`test_confirm_on_historical_thread_in_new_project_group_opens_real_thread`
-
`test_unarchive_into_inactive_existing_workspace_does_not_leave_active_draft`
-
`test_unarchive_after_removing_parent_project_group_restores_real_thread`
- `test_pending_thread_activation_suppresses_reconcile_draft_creation`
Focused test runs:
- `cargo test -p sidebar activate_archived_thread -- --nocapture`
- `cargo test -p sidebar unarchive -- --nocapture`
- `cargo test -p sidebar archive_last_thread_on_linked_worktree --
--nocapture`
- `cargo test -p sidebar
test_confirm_on_historical_thread_in_new_project_group_opens_real_thread
-- --nocapture`
- `cargo test -p sidebar
test_unarchive_into_inactive_existing_workspace_does_not_leave_active_draft
-- --nocapture`
- `cargo test -p sidebar
test_unarchive_after_removing_parent_project_group_restores_real_thread
-- --nocapture`
- `cargo test -p sidebar
test_pending_thread_activation_suppresses_reconcile_draft_creation --
--nocapture`
## Test fix
Fixed flaky `test_backfill_sets_kvp_flag` — added per-App `AppDatabase`
isolation in `setup_backfill_test` so backfill tests no longer share a
static in-memory DB.
Release Notes:
- N/A
---------
Co-authored-by: Mikayla Maki <mikayla.c.maki@gmail.com>
Co-authored-by: Mikayla Maki <mikayla@zed.dev>
When clicking a draft placeholder ("New Agent Thread" tab) or confirming
one via keyboard, the agent panel now always opens. Previously it only
focused the panel if it was already visible.
Regular threads and drafts with IDs already opened the panel through
`activate_draft` and `load_agent_thread_in_workspace`. The project
header click continues to activate the workspace without forcing the
panel open.
cc @danilo-leal
Release Notes:
- N/A
When toggling fast mode, the setting is now written to `settings.json`
under `agent.default_model.speed`, so new threads start with the same
speed. This follows the same pattern as `enable_thinking` and `effort`.
The `speed` field uses the existing `Speed` enum (`"fast"` /
`"standard"`) rather than a boolean, to leave room for future speed
tiers.
Example settings:
```json
{
"agent": {
"default_model": {
"provider": "zed.dev",
"model": "claude-sonnet-4",
"speed": "fast"
}
}
}
```
cc @benbrandt
Release Notes:
- N/A
---------
Co-authored-by: Ben Brandt <benjamin.j.brandt@gmail.com>
When a subagent was spawned while the parent had runtime-modified
settings (fast mode, thinking, etc.), the subagent would start with
defaults from `AgentSettings` instead of inheriting the parent's current
values.
This adds `inherit_parent_settings` as a single place in
`Thread::new_subagent` that copies all runtime-mutable settings from the
parent:
- `speed`
- `thinking_enabled`
- `thinking_effort`
- `summarization_model`
- `profile_id`
These are the same properties that `set_*` methods already propagate to
`running_subagents` — this fix ensures they're also inherited at
creation time.
cc @benbrandt
Release Notes:
- N/A
Extract `SyntaxTheme` into its own lightweight crate so that downstream
consumers can use syntax highlighting colors without pulling in the full
`theme` crate and its transitive dependencies.
## Changes
**Commit 1 — Extract SyntaxTheme into its own crate**
Move `SyntaxTheme`, `SyntaxThemeSettings`, `HighlightStyle`, and
supporting types from `theme/src/styles/syntax.rs` into a new
`syntax_theme` crate that depends only on `gpui`. The `theme` crate
re-exports everything for backward compatibility — no call-site changes
needed.
**Commit 2 — Add `bundled-themes` feature with One Dark**
Add an optional `bundled-themes` feature that bundles `one_dark()`, a
`SyntaxTheme` loaded from the existing One Dark JSON theme file. This
lets consumers get a usable syntax theme without depending on the full
theme machinery.
Release Notes:
- N/A
This extracts a `language_core` crate from the existing `language`
crate, and creates a `grammars` data crate. The goal is to separate
tree-sitter grammar infrastructure, language configuration, and LSP
adapter types from the heavier buffer/editor integration layer in
`language`.
## Motivation
The `language` crate pulls in `text`, `theme`, `settings`, `rpc`,
`task`, `fs`, `clock`, `sum_tree`, and `fuzzy` — all of which are needed
for buffer integration (`Buffer`, `SyntaxMap`, `Outline`,
`DiagnosticSet`) but not for grammar parsing or language configuration.
Extracting the core types lets downstream consumers depend on
`language_core` without pulling in the full integration stack.
## Dependency graph after extraction
```
language_core ← gpui, lsp, tree-sitter, util, collections
grammars ← language_core, rust_embed, tree-sitter-{rust,python,...}
language ← language_core, text, theme, settings, rpc, task, fs, ...
languages ← language, grammars
```
## What moved to `language_core`
- `Grammar`, `GrammarId`, and all query config/builder types
- `LanguageConfig`, `LanguageMatcher`, bracket/comment/indent config
types
- `HighlightMap`, `HighlightId` (theme-dependent free functions
`highlight_style` and `highlight_name` stay in `language`)
- `LanguageName`, `LanguageId`
- `LanguageQueries`, `QUERY_FILENAME_PREFIXES`
- `CodeLabel`, `CodeLabelBuilder`, `Symbol`
- `Diagnostic`, `DiagnosticSourceKind`
- `Toolchain`, `ToolchainScope`, `ToolchainList`, `ToolchainMetadata`
- `ManifestName`
- `SoftWrap`
- LSP data types: `BinaryStatus`, `ServerHealth`,
`LanguageServerStatusUpdate`, `PromptResponseContext`, `ToLspPosition`
## What stays in `language`
- `Buffer`, `BufferSnapshot`, `SyntaxMap`, `Outline`, `DiagnosticSet`,
`LanguageScope`
- `LspAdapter`, `CachedLspAdapter`, `LspAdapterDelegate` (reference
`Arc<Language>` and `WorktreeId`)
- `ToolchainLister`, `LanguageToolchainStore` (reference `task` and
`settings` types)
- `ManifestQuery`, `ManifestProvider`, `ManifestDelegate` (reference
`WorktreeId`)
- Parser/query cursor pools, `PLAIN_TEXT`, point conversion functions
## What the `grammars` crate provides
- Embedded `.scm` query files and `config.toml` files for all built-in
languages (via `rust_embed`)
- `load_queries(name)`, `load_config(name)`,
`load_config_for_feature(name, grammars_loaded)`, and `get_file(path)`
functions
- `native_grammars()` for tree-sitter grammar registration (behind
`load-grammars` feature)
## Pre-cleanup (also in this PR)
- Removed unused `Option<&Buffer>` from
`LspAdapter::process_diagnostics`
- Removed unused `&App` from `LspAdapter::retain_old_diagnostic`
- Removed `fs: &dyn Fs` from `ToolchainLister` trait methods
(`PythonToolchainProvider` captures `fs` at construction time instead)
- Moved `Diagnostic`/`DiagnosticSourceKind` out of `buffer.rs` into
their own module
## Backward compatibility
The `language` crate re-exports everything from `language_core`, so
existing `use language::Grammar` (etc.) continues to work unchanged. The
only downstream change required is importing `CodeLabelExt` where
`.fallback_for_completion()` is called on the now-foreign `CodeLabel`
type.
Release Notes:
- N/A
---------
Co-authored-by: Agus Zubiaga <agus@zed.dev>
Co-authored-by: Tom Houlé <tom@tomhoule.com>
## Motivation
This PR unifies the async execution infrastructure between GPUI and
other components that depend on the `scheduler` crate (such as our cloud
codebase). By having a scheduler that lives independently of GPUI, we
can enable deterministic testing across the entire stack - testing GPUI
applications alongside cloud services with a single, unified scheduler.
## Summary
This PR completes the integration of the `scheduler` crate into GPUI,
unifying async execution and enabling deterministic testing of GPUI
combined with other components that depend on the scheduler crate.
## Key Changes
### Scheduler Integration (Phases 1-5, previously completed)
- `TestDispatcher` now delegates to `TestScheduler` for timing, clock,
RNG, and task scheduling
- `PlatformScheduler` implements the `Scheduler` trait for production
use
- GPUI executors wrap scheduler executors, selecting `TestScheduler` or
`PlatformScheduler` based on environment
- Unified blocking logic via `Scheduler::block()`
### Dead Code Cleanup
- Deleted orphaned `crates/gpui/src/platform/platform_scheduler.rs`
(older incompatible version)
## Intentional Removals
### `spawn_labeled` and `deprioritize` removed
The `TaskLabel` system (`spawn_labeled`, `deprioritize`) was removed
during this integration. It was only used in a few places for test
ordering control.
cc @maxbrunsfeld @as-cii - The new priority-weighted scheduling in
`TestScheduler` provides similar functionality through
`Priority::High/Medium/Low`. If `deprioritize` is important for specific
test scenarios, we could add it back to the scheduler crate. Let me know
if this is blocking anything.
### `start_waiting` / `finish_waiting` debug methods removed
Replaced by `TracingWaker` in `TestScheduler` - run tests with
`PENDING_TRACES=1` to see backtraces of pending futures when parking is
forbidden.
### Realtime Priority removed
The realtime priority feature was unused in the codebase. I'd prefer to
reintroduce it when we have an actual use case, as the implementation
(bounded channel with capacity 1) could potentially block the main
thread. Having a real use case will help us validate the design.
## Testing
- All GPUI tests pass
- All scheduler tests pass
- Clippy clean
## Architecture
```
┌─────────────────────────────────────────────────────────────┐
│ GPUI │
│ ┌──────────────────────┐ ┌────────────────────────────┐ │
│ │ gpui::Background- │ │ gpui::ForegroundExecutor │ │
│ │ Executor │ │ - wraps scheduler:: │ │
│ │ - scheduler: Arc< │ │ ForegroundExecutor │ │
│ │ dyn Scheduler> │ └────────────┬───────────────┘ │
│ └──────────┬───────────┘ │ │
│ │ │ │
│ └──────────┬──────────────────┘ │
│ ▼ │
│ ┌───────────────────────┐ │
│ │ Arc<dyn Scheduler> │ │
│ └───────────┬───────────┘ │
│ ┌──────────────┴──────────────┐ │
│ ▼ ▼ │
│ ┌──────────────────┐ ┌────────────────────┐ │
│ │ PlatformScheduler│ │ TestScheduler │ │
│ │ (production) │ │ (deterministic) │ │
│ └──────────────────┘ └────────────────────┘ │
└─────────────────────────────────────────────────────────────┘
```
Release Notes:
- N/A
---------
Co-authored-by: Antonio Scandurra <me@as-cii.com>
Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>
Co-authored-by: Yara <git@yara.blue>
Co-authored-by: Zed Zippy <234243425+zed-zippy[bot]@users.noreply.github.com>
This brings the terminal element's viewport culling in line with the
editor optimization in PR #44995 and the fix in PR #45077.
## Problem
When a terminal is inside a scrollable container (e.g., the Agent Panel
thread view), it would render ALL cells during prepaint, even when the
terminal was entirely outside the viewport. This caused unnecessary CPU
usage when multiple terminal tool outputs existed in the Agent Panel.
## Solution
Calculate the intersection of the terminal's bounds with the current
content_mask (the visible viewport after all parent clipping). If the
intersection has zero area, skip all cell processing entirely.
### Three code paths
1. **Offscreen** (`intersection.size <= 0`): Early exit, process 0 cells
2. **Fully visible** (`intersection == bounds`): Fast path, stream cells
directly (no allocation)
3. **Partially clipped**: Group cells by line, skip/take visible rows
only
### Key insight: filter by screen position, not buffer coordinates
The previous approach tried to filter cells by `cell.point.line`
(terminal buffer coordinates), which breaks in Scrollable mode where
cells can have negative line numbers for scrollback history.
The new approach filters by **screen position** using
`chunk_by(line).skip(N).take(M)`, which works regardless of the actual
line numbers because we're filtering on enumerated line group index.
## Testing
Added comprehensive unit tests for:
- Screen-position filtering with positive lines (Inline mode)
- Screen-position filtering with negative lines (Scrollable mode with
scrollback)
- Edge cases (skip all, positioning math)
- Unified filtering works for both modes
Manually verified:
- Terminal fully visible (no clipping) ✓
- Terminal clipped from top/bottom ✓
- Terminal completely outside viewport ✓
- Scrollable terminals with scrollback history ✓
- Selection/interaction still works ✓
Release Notes:
- Improved Agent Panel performance when terminals are scrolled
offscreen.
/cc @as-cii
Release Notes:
- Fixed spurious "no checkpoint" error in agent panel
---
## Summary
`update_last_checkpoint` would call `last_user_message()` twice - once
at the start to capture the checkpoint, and again in an async closure
after the checkpoint comparison completed. If a new user message without
a checkpoint was added between these two calls, the second call would
find the new message and fail with "no checkpoint".
## Fix
Capture the user message ID at the start and use `user_message_mut(&id)`
in the async closure to find the specific message.
cc @mikayla-maki
## Problem
When users paste or drag large images into the agent panel, the encoded
payload can exceed upstream provider limits (e.g., Anthropic's 5MB
per-image limit), causing API errors.
## Solution
Enforce a default 5MB limit on encoded PNG bytes in
`LanguageModelImage::from_image`:
1. Apply existing Anthropic dimension limits first (1568px max in either
dimension)
2. Iteratively downscale by ~15% per pass until the encoded PNG is under
5MB
3. Return `None` if the image can't be shrunk within 8 passes
(fail-safe)
The limit is enforced at the `LanguageModelImage` conversion layer,
which is the choke point for all image ingestion paths (agent panel
paste/drag, file mentions, text threads, etc.).
## Future Work
The 5MB limit is a conservative default. Provider-specific limits can be
introduced later by adding a `from_image_with_constraints` API.
## Testing
Added a regression test that:
1. Generates a noisy 4096x4096 PNG (guaranteed >5MB)
2. Converts it via `LanguageModelImage::from_image`
3. Asserts the result is ≤5MB and was actually downscaled
---
**Note:** This PR builds on #45312 (prompt store fail-open fix). Please
merge that first.
cc @rtfeldman
---------
Co-authored-by: Zed Zippy <234243425+zed-zippy[bot]@users.noreply.github.com>
Generated by AI.
`DocumentTypes.plist` declares `CFBundleTypeIconFile` as `Document` for
Zed’s document types, but the macOS bundle did not include
`Contents/Resources/Document.icns`, causing Finder to fall back to
generic icons.
This PR:
- Adds `crates/zed/resources/Document.icns` as a placeholder document
icon (currently derived from the app icon).
- Updates `script/bundle-mac` to copy it into the `.app` at
`Contents/Resources/Document.icns` during bundling.
- Adds `script/verify-macos-document-icon` for one-command validation.
## How to test (CLI)
1. Build a debug bundle:
- `./script/bundle-mac -d aarch64-apple-darwin`
2. Verify the bundle contains the referenced icon:
- `./script/verify-macos-document-icon
"target/aarch64-apple-darwin/debug/bundle/osx/Zed Dev.app"`
## Optional visual validation in Finder
- Pick a file (e.g. `.rs`), Get Info → Open with: Zed Dev → Change
All...
- Restart Finder: `killall Finder` (or log out/in)
@JosephTLyons — would you mind running the steps above and confirming
Finder shows Zed’s icon for source files after "Change All" + Finder
restart?
@danilo-leal — this PR ships a placeholder `Document.icns`. When the
real document icon is ready, replace
`crates/zed/resources/Document.icns` and the bundling script will
include it automatically.
Closes#44403.
Release Notes:
- TODO
---------
Co-authored-by: Matt Miller <mattrx@gmail.com>
Closes#32951
## Summary
When an agent notification was shown and the `AcpThreadView` was dropped
(e.g., by closing the project window or navigating to a new thread), the
notification would become orphaned and undismissable because the
subscriptions handling dismiss events were dropped along with the thread
view.
## Fix
Added an `on_release` callback that closes all notification windows when
the thread view is dropped. This ensures notifications are always
cleaned up properly.
## Testing
Added `test_notification_closed_when_thread_view_dropped` to verify
notifications are closed when the thread view is dropped.
Release Notes:
- Fixed agent notification getting stuck and becoming undismissable when
the project window is closed or when navigating to a new thread
This reverts commit f576ec7a15 (#44995).
The optimization introduced a regression that causes the main thread to
hang for **100+ seconds** in certain scenarios, requiring a force quit
to recover.
## Analysis from spindump
When a large `AutoHeight` editor is displayed inside a `List` (e.g.,
Agent Panel thread view), the clipping calculation can produce invalid
row ranges:
1. `visible_bounds` from `window.content_mask().bounds` represents the
window's content mask, not the intersection with the editor
2. When the editor is partially scrolled out of view,
`clipped_top_in_lines` becomes extremely large
3. This causes `start_row` to be computed as an astronomically high
value
4. `blocks_in_range(start_row..end_row)` then spends excessive time in
`Cursor::search_forward` iterating through the block tree
The spindump showed **~46% of samples** (459/1001 over 10+ seconds)
stuck in `BlockSnapshot::blocks_in_range()`, specifically in cursor
iteration.
### Heaviest stack trace
```
EditorElement::prepaint
└─ blocks_in_range + 236
└─ Cursor::search_forward (459 samples)
```
## Symptoms
- Main thread unresponsive for 33-113 seconds before sampling even began
- UI completely frozen
- High CPU usage on main thread (10+ seconds of CPU time in the sample)
- Force quit required to recover
## Path forward
The original optimization goal (reducing line layout work for clipped
editors) is valid, but the implementation needs to:
1. Correctly calculate the **intersection** of editor bounds with the
visible viewport
2. Ensure row calculations stay within valid ranges (clamped to
`max_row`)
3. Handle edge cases where the editor is completely outside the visible
bounds
Release Notes:
- Fixed a hang that could occur when viewing large diffs in the Agent
Panel
This re-introduces the `save_file` and `restore_file_from_disk` agent
tools that were reverted in #44949.
I pushed that original PR without trying it just to get the build off my
machine, but I had missed a step: the tools weren't added to the default
profile settings in `default.json`, so they were never enabled even
though the code was present.
## Changes
- Add `save_file` and `restore_file_from_disk` to the "write" profile in
`default.json`
- Add `Thread::has_tool()` method to check tool availability at runtime
- Make `edit_file_tool`'s dirty buffer error message conditional on
whether `save_file`/`restore_file_from_disk` tools are available (so the
agent gets appropriate guidance based on what tools it actually has)
- Update test to match new conditional error message behavior
Release Notes:
- Added `save_file` and `restore_file_from_disk` agent tools to handle
dirty buffers when editing files
Release Notes:
- Added `save_file` and `restore_file_from_disk` tools to the agent,
allowing it to resolve dirty buffer conflicts when editing files. When
the agent encounters a file with unsaved changes, it will now ask
whether you want to keep or discard those changes before proceeding.
Adds an optional `timeout_ms` parameter to the terminal tool that allows
bounding the runtime of shell commands. When the timeout expires, the
running terminal task is killed and the tool returns with the partial
output captured so far.
## Summary
This PR adds the ability for the agent to specify a maximum runtime when
invoking the terminal tool. This helps prevent indefinite hangs when
running commands that might wait for network, user prompts, or long
builds/tests.
## Changes
- Add `timeout_ms` field to `TerminalToolInput` schema
- Extend `TerminalHandle` trait with `kill()` method
- Implement `kill()` for `AcpTerminalHandle` and `EvalTerminalHandle`
- Race terminal exit against timeout, killing on expiry
- Update system prompt to recommend using timeouts for long-running
commands
- Add test for timeout behavior
- Update `.rules` to document GPUI executor timers for tests
## Testing
- Added `test_terminal_tool_timeout_kills_handle` which verifies that
when a timeout is specified and expires, the terminal handle is killed
and the tool returns with partial output.
- All existing agent tests pass.
Release Notes:
- agent: Added optional `timeout_ms` parameter to the terminal tool,
allowing the agent to bound command runtime and prevent indefinite hangs
The test `test_collaborating_with_completion` has a latent race
condition that hasn't manifested on CI yet but could cause hangs with
certain task orderings.
## The Bug
Commit `f3d9fc0013` set up LSP request handlers AFTER typing the trigger
character:
```rust
// Type trigger first - spawns async tasks to send completion request
editor_b.update_in(cx_b, |editor, window, cx| {
editor.handle_input(".", window, cx);
});
// THEN set up handlers (race condition!)
fake_language_server
.set_request_handler::<lsp::request::Completion, _, _>(...)
.next().await.unwrap(); // Waits for handler to receive a request
```
Whether this works depends on task scheduling order, which varies by
seed. If the completion request is processed before the handler is
registered, the request goes to `on_unhandled_notification` which claims
to handle it but sends no response, causing a hang.
## Changes
- Move handler setup BEFORE typing the trigger character
- Make `TestDispatcher::spawn_realtime` panic to prevent future
non-determinism from real OS threads
- Add `execution_hash()` and `execution_count()` to TestDispatcher for
debugging
- Add `DEBUG_SCHEDULER=1` logging for task execution tracing
- Document the investigation in `situation.md`
cc @localcc @SomeoneToIgnore (authors of related commits)
Release Notes:
- N/A
---------
Co-authored-by: Kirill Bulatov <kirill@zed.dev>
This adds implementations of `std::iter::Sum` for `Pixels`, allowing the
use of `.sum()` on iterators of `Pixels` values.
### Changes
- Implement `Sum<Pixels>` for `Pixels` (owned values)
- Implement `Sum<&Pixels>` for `Pixels` (references)
This enables ergonomic patterns like:
```rust
let total: Pixels = pixel_values.iter().sum();
```
This allows ITERATIONS and SEED environment variables to override the
hard coded values during testing.
cc @ConradIrwin @as-cii
Release Notes:
- N/A
---------
Co-authored-by: Conrad Irwin <conrad.irwin@gmail.com>
Extracts and cleans up GPUI's scheduler code into a new `scheduler`
crate, making it pluggable by external runtimes. This will enable
deterministic integration testing with cloud components by providing a
unified test scheduler across Zed and backend code. In Zed, it will
replace the existing GPUI scheduler for consistent async task management
across platforms.
## Changes
- **Core Implementation**: `TestScheduler` with seed-based
randomization, session tracking (`SessionId`), and foreground/background
task separation for reproducible testing.
- **Executors**: `ForegroundExecutor` (!Send, thread-local) and
`BackgroundExecutor` (Send, with blocking/timeout support) as
GPUI-compatible wrappers.
- **Clock and Timer**: Controllable `TestClock` and future-based `Timer`
for time-sensitive tests.
- **Testing APIs**: `once()`, `with_seed()`, and `many()` methods for
configurable test runs.
- **Dependencies**: Added `async-task`, `chrono`, `futures`, etc., with
updates to `Cargo.toml` and lock file.
## Benefits
- **Integration Testing**: Facilitates reliable async tests involving
cloud sessions, reducing flakiness via deterministic execution.
- **Pluggability**: Trait-based design (`Scheduler`) allows easy
integration into non-GPUI runtimes while maintaining GPUI compatibility.
- **Cleanup**: Refactors GPUI scheduler logic for clarity, correctness
(no `unwrap()`, proper error handling), and extensibility.
Follows Rust guidelines; run `./script/clippy` for verification.
- [x] Define and test a core scheduler that we think can power our cloud
code and GPUI
- [ ] Replace GPUI's scheduler
Release Notes:
- N/A
---------
Co-authored-by: Antonio Scandurra <me@as-cii.com>
Release Notes:
- Added channel reordering for administrators (use `cmd-up` and
`cmd-down` on macOS or `ctrl-up` `ctrl-down` on Linux to move channels
up or down within their parent)
## Summary
This PR introduces the ability for channel administrators to reorder
channels within their parent context, providing better organizational
control over channel hierarchies. Users can now move channels up or down
relative to their siblings using keyboard shortcuts.
## Problem
Previously, channels were displayed in alphabetical order with no way to
customize their arrangement. This made it difficult for teams to
organize channels in a logical order that reflected their workflow or
importance, forcing users to prefix channel names with numbers or
special characters as a workaround.
## Solution
The implementation adds a persistent `channel_order` field to channels
that determines their display order within their parent. Channels with
the same parent are sorted by this field rather than alphabetically.
## Implementation Details
### Database Schema
Added a new column and index to support efficient ordering:
```sql
-- crates/collab/migrations/20250530175450_add_channel_order.sql
ALTER TABLE channels ADD COLUMN channel_order INTEGER NOT NULL DEFAULT 1;
CREATE INDEX CONCURRENTLY "index_channels_on_parent_path_and_order" ON "channels" ("parent_path", "channel_order");
```
### RPC Protocol
Extended the channel proto with ordering support:
```proto
// crates/proto/proto/channel.proto
message Channel {
uint64 id = 1;
string name = 2;
ChannelVisibility visibility = 3;
int32 channel_order = 4;
repeated uint64 parent_path = 5;
}
message ReorderChannel {
uint64 channel_id = 1;
enum Direction {
Up = 0;
Down = 1;
}
Direction direction = 2;
}
```
### Server-side Logic
The reordering is handled by swapping `channel_order` values between
adjacent channels:
```rust
// crates/collab/src/db/queries/channels.rs
pub async fn reorder_channel(
&self,
channel_id: ChannelId,
direction: proto::reorder_channel::Direction,
user_id: UserId,
) -> Result<Vec<Channel>> {
// Find the sibling channel to swap with
let sibling_channel = match direction {
proto::reorder_channel::Direction::Up => {
// Find channel with highest order less than current
channel::Entity::find()
.filter(
channel::Column::ParentPath
.eq(&channel.parent_path)
.and(channel::Column::ChannelOrder.lt(channel.channel_order)),
)
.order_by_desc(channel::Column::ChannelOrder)
.one(&*tx)
.await?
}
// Similar logic for Down...
};
// Swap the channel_order values
let temp_order = channel.channel_order;
channel.channel_order = sibling_channel.channel_order;
sibling_channel.channel_order = temp_order;
}
```
### Client-side Sorting
Optimized the sorting algorithm to avoid O(n²) complexity:
```rust
// crates/collab/src/db/queries/channels.rs
// Pre-compute sort keys for efficient O(n log n) sorting
let mut channels_with_keys: Vec<(Vec<i32>, Channel)> = channels
.into_iter()
.map(|channel| {
let mut sort_key = Vec::with_capacity(channel.parent_path.len() + 1);
// Build sort key from parent path orders
for parent_id in &channel.parent_path {
sort_key.push(channel_order_map.get(parent_id).copied().unwrap_or(i32::MAX));
}
sort_key.push(channel.channel_order);
(sort_key, channel)
})
.collect();
channels_with_keys.sort_by(|a, b| a.0.cmp(&b.0));
```
### User Interface
Added keyboard shortcuts and proper context handling:
```json
// assets/keymaps/default-macos.json
{
"context": "CollabPanel && not_editing",
"bindings": {
"cmd-up": "collab_panel::MoveChannelUp",
"cmd-down": "collab_panel::MoveChannelDown"
}
}
```
The CollabPanel now properly sets context to distinguish between editing
and navigation modes:
```rust
// crates/collab_ui/src/collab_panel.rs
fn dispatch_context(&self, window: &Window, cx: &Context<Self>) -> KeyContext {
let mut dispatch_context = KeyContext::new_with_defaults();
dispatch_context.add("CollabPanel");
dispatch_context.add("menu");
let identifier = if self.channel_name_editor.focus_handle(cx).is_focused(window) {
"editing"
} else {
"not_editing"
};
dispatch_context.add(identifier);
dispatch_context
}
```
## Testing
Comprehensive tests were added to verify:
- Basic reordering functionality (up/down movement)
- Boundary conditions (first/last channels)
- Permission checks (non-admins cannot reorder)
- Ordering persistence across server restarts
- Correct broadcasting of changes to channel members
## Migration Strategy
Existing channels are assigned initial `channel_order` values based on
their current alphabetical sorting to maintain the familiar order users
expect:
```sql
UPDATE channels
SET channel_order = (
SELECT ROW_NUMBER() OVER (
PARTITION BY parent_path
ORDER BY name, id
)
FROM channels c2
WHERE c2.id = channels.id
);
```
## Future Enhancements
While this PR provides basic reordering functionality, potential future
improvements could include:
- Drag-and-drop reordering in the UI
- Bulk reordering operations
- Custom sorting strategies (by activity, creation date, etc.)
## Checklist
- [x] Database migration included
- [x] Tests added for new functionality
- [x] Keybindings work on macOS and Linux
- [x] Permissions properly enforced
- [x] Error handling implemented throughout
- [x] Manual testing completed
- [x] Documentation updated
---------
Co-authored-by: Mikayla Maki <mikayla.c.maki@gmail.com>
You can set `agent_font_size` as a top-level settings key. You can also
use `zed::IncreaseBufferFontSize` and `zed::DecreaseBufferFontSize` and
`zed::ResetBufferFontSize` the agent panel is focused via the standard
bindings to adjust the agent font size. In the future, it might make
sense to rename these actions to be more general since "buffer" is now a
bit of a misnomer. 🍐'd with @mikayla-maki
Release Notes:
- N/A
---------
Co-authored-by: Mikayla Maki <mikayla.c.maki@gmail.com>
I am currently setting the font size corrrectly by using a custom
EditorStyle and building an element. However I need to use the same
properties as a normal editor for everything but font size.
Release Notes:
- N/A
This is to enable alternative streaming solutions at the application
layer. I'm not sure we really should have performed parsing of the input
at this layer. Either way I want to experiment with streaming approaches
in a separate crate on a branch, and this will help.
/cc @maxdeviant @bennetbo @rtfeldman
Closes #ISSUE
Release Notes:
- N/A
The old one wasn't linking, and
https://github.com/zed-industries/zed/pull/29081 has a bunch of merge
conflicts. Wanted to start simple/small.
## Todo
* [x] Remove low-signal examples
* [x] Make the eval run on a cron, on main, and on any PR with the
`run-eval` label
* [x] Noise in logs about failure to write settings
```
[2025-04-21T20:45:04Z ERROR settings] Failed to write settings to file
"/home/runner/.config/zed/settings.json"
Caused by:
No such file or directory (os error 2) at path
"/home/runner/.config/zed/.tmpLewFEs"
```
* [x] `Agentic loop stalled`
(https://github.com/zed-industries/zed/actions/runs/14581044243/job/40897622894)
* [x] Make sure that events are recorded in snowflake
* [ ] Change judge criteria to be more explicit about meanings of scores
Release Notes:
- N/A
---------
Co-authored-by: Antonio Scandurra <me@as-cii.com>
Co-authored-by: Agus Zubiaga <hi@aguz.me>
Co-authored-by: Max Brunsfeld <maxbrunsfeld@gmail.com>
Co-authored-by: Thomas Mickley-Doyle <tmickleydoyle@gmail.com>
This PR renames the `regex_search` tool to `grep` because I think it
conveys more meaning to the model, the idea of searching the filesystem
with a regular expression. It's also one word and the model seems to be
using it effectively after some additional prompt tuning.
It also takes an include pattern to filter on the specific files we try
to search. I'd like to encourage the model to scope its searches more
aggressively, as in my testing, I'm only seeing it filter on file
extension.
Release Notes:
- N/A
Previously, we were including the dummy text "Here are the tool
results." whenever reporting tool call results. I'm worried this is
adding noise and confusing the model, because the user didn't actually
say anything. This inserts an empty message to be populated later. My
preference would be something less stateful, where tool results (or
batches of them requested simultaneously) would be sent to the model as
soon as they were ready, without bothering to do this message
association dance. But for now, this seems to work.
Release Notes:
- N/A
Now that we've established a proper eval in tree, this PR is reboots of
our agent loop back to a set of minimal tools and simpler prompts. We
should aim to get this branch feeling subjectively competitive with
what's on main and then merge it, and build from there.
Let's invest in our eval and use it to drive better performance of the
agent loop. How you can help: Pick an example, and then make the outcome
faster or better. It's fine to even use your own subjective judgment, as
our evaluation criteria likely need tuning as well at this point. Focus
on making the agent work better in your own subjective experience first.
Let's focus on simple/practical improvements to make this thing work
better, then determine how we can craft our judgment criteria to lock
those improvements in.
Release Notes:
- N/A
---------
Co-authored-by: Max <max@zed.dev>
Co-authored-by: Antonio <antonio@zed.dev>
Co-authored-by: Agus <agus@zed.dev>
Co-authored-by: Richard <richard@zed.dev>
Co-authored-by: Max Brunsfeld <maxbrunsfeld@gmail.com>
Co-authored-by: Antonio Scandurra <me@as-cii.com>
Co-authored-by: Michael Sloan <mgsloan@gmail.com>
Previously, when comparing a block with an empty range to an empty query
range in non-inclusive mode, our binary search logic could end up
computing an inverted range, causing a panic.
This commit adds special casing when comparing empty blocks with empty
ranges.
cc @as-cii: I'm realizing that the approach to searching for the
intersecting replacement blocks makes some invalid assumptions about the
ordering of replace decorations. They aren't ordered at all by their end
range. @maxbrunsfeld and I are wondering if long term, we should remove
replace decorations and find another solution for folding buffers in
multi buffers.
Release Notes:
- Fixed an occasional panic that would occur when navigating to the next
change hunk with a pending inline transformation present.
Co-authored-by: Peter Tripp <petertripp@gmail.com>
We think this could fix issues around view invalidation during focus
handling.
I want to run CI on this and see.
cc @mikayla-maki @maxbrunsfeld
Release Notes:
- N/A
There's still a bit more work to do on this, but this PR is compiling
(with warnings) after eliminating the key types. When the tasks below
are complete, this will be the new narrative for GPUI:
- `Entity<T>` - This replaces `View<T>`/`Model<T>`. It represents a unit
of state, and if `T` implements `Render`, then `Entity<T>` implements
`Element`.
- `&mut App` This replaces `AppContext` and represents the app.
- `&mut Context<T>` This replaces `ModelContext` and derefs to `App`. It
is provided by the framework when updating an entity.
- `&mut Window` Broken out of `&mut WindowContext` which no longer
exists. Every method that once took `&mut WindowContext` now takes `&mut
Window, &mut App` and every method that took `&mut ViewContext<T>` now
takes `&mut Window, &mut Context<T>`
Not pictured here are the two other failed attempts. It's been quite a
month!
Tasks:
- [x] Remove `View`, `ViewContext`, `WindowContext` and thread through
`Window`
- [x] [@cole-miller @mikayla-maki] Redraw window when entities change
- [x] [@cole-miller @mikayla-maki] Get examples and Zed running
- [x] [@cole-miller @mikayla-maki] Fix Zed rendering
- [x] [@mikayla-maki] Fix todo! macros and comments
- [x] Fix a bug where the editor would not be redrawn because of view
caching
- [x] remove publicness window.notify() and replace with
`AppContext::notify`
- [x] remove `observe_new_window_models`, replace with
`observe_new_models` with an optional window
- [x] Fix a bug where the project panel would not be redrawn because of
the wrong refresh() call being used
- [x] Fix the tests
- [x] Fix warnings by eliminating `Window` params or using `_`
- [x] Fix conflicts
- [x] Simplify generic code where possible
- [x] Rename types
- [ ] Update docs
### issues post merge
- [x] Issues switching between normal and insert mode
- [x] Assistant re-rendering failure
- [x] Vim test failures
- [x] Mac build issue
Release Notes:
- N/A
---------
Co-authored-by: Antonio Scandurra <me@as-cii.com>
Co-authored-by: Cole Miller <cole@zed.dev>
Co-authored-by: Mikayla <mikayla@zed.dev>
Co-authored-by: Joseph <joseph@zed.dev>
Co-authored-by: max <max@zed.dev>
Co-authored-by: Michael Sloan <michael@zed.dev>
Co-authored-by: Mikayla Maki <mikaylamaki@Mikaylas-MacBook-Pro.local>
Co-authored-by: Mikayla <mikayla.c.maki@gmail.com>
Co-authored-by: joão <joao@zed.dev>
Previously, each window stored its own collection of focus handles. This
meant that to create a focus handle, you needed to have access to a
Window. I'm working on a simplification to gpui's context types that
removes `WindowContext` and `ViewContext` in favor of passing a window
reference explicitly when rendering or handling events. You'll still
need a window to manipulate focus, but it will be helpful to be able to
create focus handles without a window.
cc @mgsloan
Release Notes:
- N/A
@bennetbo @as-cii @mrnugget I'm really not liking the hints about AI on
every line. It feels too distracting to me and damaging to the user
experience. I'm wondering if we can hide them and work with design for
other ideas. Or at least talk it through.
Release Notes:
- N/A
This will allow us to use the events table directly in Amplitude, which
lets us use the newer event ingestion flow that detects changes to the
table. Otherwise we'll need a transformation.
I think Amplitude's API is probably a pretty good example to follow for
the raw event schema, even if we don't end up using their product. They
also recommend a "Noun Verbed" format for naming events, so I think we
should go with this. This will help us be consistent and encourage the
author of events to think more clearly about what event they're
reporting.
cc @ConradIrwin
Release Notes:
- N/A
Previously, we were reaching in and using the context_stack on the dispatch tree, which was incorrect.
/cc @as-cii
/cc @ConradIrwin
Release Notes:
- N/A
---------
Co-authored-by: Michael Sloan <michael@zed.dev>
Hopefully, this will help people understand how easy it is to add
context to an inline transformation.
@as-cii @maxdeviant @rtfeldman could somebody update this to display the
actual correct key bindings and ship it. I have them hard coded for now.
Release Notes:
- Updated placeholder text with key bindings to focus context panel and
navigate history.
---------
Co-authored-by: Richard Feldman <oss@rtfeldman.com>
When we insert before some text and then update that same text, we need
to preserve and concatenate the new text associated with both
operations.
Release Notes:
- N/A
This allows us to detect the language from the extension if we use paths
in fenced code blocks.
Release Notes:
- You can now use file paths ending in a language-specific file
extension at the start of markdown code blocks.
This commit adds a custom icon for Anthropic hosted models.
- Adding a new SVG icon for Anthropic hosted models.
- The new icon is located at: `assets/icons/ai_anthropic_hosted.svg`
- Updating the LanguageModel trait to include an optional icon method
- Implementing the icon method for CloudModel to return the custom icon
for Anthropic hosted models
- Updating the UI components to use the model-specific icon when
available
- Adding a new IconName variant for the Anthropic hosted icon
We should change the non-hosted icon in some small way to distinguish it
from the hosted version. I duplicated the path for now so we can
hopefully add it for the next release.
Release Notes:
- N/A
When launching Zed from the CLI via `cargo run`, we'll always prompt
load templates from the repo.
This restores behavior that I reverted last night in #16403.
Also, I've improved the `script/prompts link/unlink` workflow for
overriding prompts of your production copy of Zed. Zed now detects when
the overrides directory is created or removed, and does the right thing.
You can link and unlink repeatedly without restarting Zed.
Release Notes:
- N/A
This PR reverts #16145 and subsequent changes.
This reverts commit 8c8e89eeb0.
We still have issues with our approach to indentation in Python
unfortunately, but this feels like a safer equilibrium than where we
were.
Release Notes:
- Returned to our previous prompt for inline assist transformations,
since recent changes were introducing issues.
I focused on cases where we're inserting doc comments or annotations
above symbols.
I added 5 new examples to the content generation prompt, covering
various scenarios:
1. Inserting documentation for a Rust struct
2. Writing docstrings for a Python class
3. Adding comments to a TypeScript method
4. Adding a derive attribute to a Rust struct
5. Adding a decorator to a Python class
These examples demonstrate how to handle different languages and common
tasks like adding documentation, attributes, and decorators.
To improve context integration, I've made the following changes:
1. Added a `transform_context_range` that includes 3 lines before and
after the transform range
2. Introduced `rewrite_section_prefix` and `rewrite_section_suffix` to
provide more context around the section being rewritten
3. Updated the prompt template to include this additional context in a
separate code snippet
Release Notes:
- Reduced instances of over-generation when inserting docs or
annotations above a symbol.
Reverts zed-industries/zed#15981
Release Notes:
- Restored the behavior of `ctrl-a` until we can separate the behavior
of the command for soft wraps and leading indentation.
This PR simplifies our approach to indentation in the inline assistant
in hopes of improving our experience for Python. We tell the model to
generate the correct indentation in the prompt, and always start
generating at the start of the line. This may fall down for less capable
models, but I want to get a solid experience on the best models and then
figure the rest out later.
Also: We now prefer `./assets/prompts` as an overrides directory when
stdout is a PTY, so you can do `cargo run` and then iterate prompts for
the current run inside the current working copy.
cc @trishume @dsp-ant
Release Notes:
- Zed now allows the model to control indentation when performing inline
transformation. We're hoping this improves the indentation experience in
Python and other indentation-sensitive languages, but it does require
more from the model.
---------
Co-authored-by: Mikayla <mikayla@zed.dev>
# HTTP Client Improvements for Extension API
This PR enhances the HTTP client functionality in the Zed extension API,
providing more control over requests and allowing for streaming
responses.
## Key Changes
1. Extended `HttpRequest` struct:
- Added `method` field to specify HTTP method
- Added `headers` field for custom headers
- Added optional `body` field for request payload
2. Introduced `HttpMethod` enum for supported HTTP methods
3. Updated `HttpResponse` struct:
- Added `headers` field to access response headers
- Changed `body` type from `String` to `Vec<u8>` for binary data support
4. Added streaming support:
- New `fetch_stream` function to get a response stream
- Introduced `HttpResponseStream` resource for chunked reading
5. Updated internal implementations to support these new features
6. Modified the Gleam extension to use the new API structure
## Motivation
These changes provide extension developers with more flexibility and
control over HTTP requests. The streaming support is particularly useful
for handling large responses efficiently or ideally streaming into the
UI.
## Testing
- [x] Updated existing tests
- [ ] Added new tests for streaming functionality
## Next Steps
- Consider adding more comprehensive examples in the documentation
- Evaluate performance impact of streaming for large responses
Please review and let me know if any adjustments are needed.
Release Notes:
- N/A
---------
Co-authored-by: Marshall Bowers <elliott.codes@gmail.com>
This PR introduces a new script for iterative development of prompt
overrides in Zed.
Just `script/prompts link` and your running Zed should start using
prompts from `zed/assets/prompts`. Use `script/prompts unlink` to undo.
You can also link with `script/prompts link --worktree` to store the
prompts to a `../zed_prompts` worktree that's a sibling of your repo, in
case you don't want to mess with your working copy. Just don't forget
about it!
Key changes:
- Add new `script/prompts` for managing prompt overrides
- Rename `prompt_templates_dir` to `prompt_overrides_dir` for clarity
- Update paths to use `~/.config/zed/prompt_overrides` instead of
`~/.config/zed/prompts/templates`
- Adjust `PromptBuilder` to use the new `prompt_overrides_dir` function
These changes simplify the process of customizing prompts and provide a
more intuitive naming convention for override-related functionality.
Release Notes:
- N/A
Copies rust files from extension_api/wit to the OUT_DIR to allow
including them from within the crate, which is supported by
rust-analyzer. This allows rust-analyzer to deal with the included
files. It doesn't currently support files outside the crate.
Release Notes:
- N/A
This PR opens workflow step editors as preview tabs and closes them upon
exiting the step if they are still in preview mode and they weren't
already open before entering the step.
Making this work was tricky, because we often edit the buffer as part of
displaying the workflow step suggestions to create empty lines where we
can generate. We undo these edits if the transformation is not applied,
but they were causing the preview to be dismissed.
After trying a few approaches, I decided to give workspace `Item`s a
`preserve_preview` method that defaults to false. When the workspace
sees an edit event for the item, it checks if the item wants to preserve
its preview. For buffers, after editing, you can call `refresh_preview`,
which sets a preview version to the current version of the buffer. Any
edits after this version will cause preview to not be preserved.
One final issue is with async auto-indent. To ensure these async edits
don't dismiss the preview, I automatically refresh the preview version
if preview was preserved prior to performing the auto-indent. The
assumption is that these are edits created by other edits, and if we
didn't want to dismiss the preview with the originating edits, then the
auto-indent edits shouldn't dismiss it either.
Release Notes:
- N/A
---------
Co-authored-by: Jason <jason@zed.dev>
This will help us as we hit issues with the /workflow and step
resolution. We can override the baked-in prompts and make tweaks, then
import our refinements back into the source tree when we're ready.
Release Notes:
- N/A
I need this to refine our prompts on the fly as I work.
Release Notes:
- Templates for prompts driving inline transformation in editors and the
terminal can now be overridden in the `~/.config/zed/prompts/templates`
directory. This is an advanced feature, and prevents you from getting
upstream changes. It's intended for use by Zed developers.
After the changes in #15536, block decorations need to be given an
explicit height if their content doesn't consume height on its own. We
missed that inline transformation deletion decorations didn't do this,
creating weird results. This fixes the issue and prevents block
decorations from ever having a zero height. That helps avoid major
weirdness, but this still a bit of a gotcha.
We need to back port this to Preview Channel (0.147.x)
Release Notes:
- N/A
---------
Co-authored-by: Antonio <antonio@zed.dev>
Built-in prompts can still be removed from the default prompt, but they
can't be edited and are automatically updated with new Zed releases.
Release Notes:
- N/A
---------
Co-authored-by: Antonio <antonio@zed.dev>
This adds support for detecting line comments in the
[Jupytext](https://jupytext.readthedocs.io/) format. When line comments
such as `# %%` is present, invoking `repl: run` will evaluate the code
between these line comments as a unit.
/cc @rgbkrk
```py
# %%
# This is my first block
print(1)
print(2)
# %%
# This is my second block
print(3)
```
Release Notes:
- N/A
---------
Co-authored-by: Antonio Scandurra <me@as-cii.com>
Co-authored-by: Antonio <antonio@zed.dev>
Co-authored-by: Thorsten <thorsten@zed.dev>
Co-authored-by: Thorsten Ball <mrnugget@gmail.com>
Also, allow proto messages to be deserialized. This is to support
translating these messages JS types in a new server implementation based
on CloudFlare durable objects.
Release Notes:
- N/A
Release Notes:
- Fixed a bug that was causing system prompts to be dropped for
Anthropic models.
@JosephTLyons @notpeter We probably need to hot-fix this as I'm pretty
sure this affects the regular anthropic provider in addition to just the
feature-flagged cloud stuff. Wouldn't mind confirming that first so we
can communicate around it. 😬
Using the file system as a database seems like it's easy, but it's
actually a real pain. I'd like to use LMDB to store the prompts locally
so we have more control. We can always add an export option, but I want
the source of truth to be somewhere other than the file system.
So far, I have a PromptStore which is global to the application and can
be initialized on startup. Then there's a `PromptLibrary` which is
intended to be the root of a new kind of Zed window. I haven't actually
seen pixels yet, but I've sketched out the basics needed to create a new
prompt, save, etc.
Still lots to figure out but the foundations of being backed by a DB and
rendering in an independent window are in place.
/cc @iamnbutler @as-cii
Release Notes:
- N/A
---------
Co-authored-by: Antonio Scandurra <me@as-cii.com>
This is still pretty raw, so I'd like to hold off on shipping it to all
users.
Release Notes:
- Hide the prompt library for non-staff until it is in a more complete
state.
We should reintroduce this as part of the prompt library.
Release Notes:
- Removed an over-eager system prompt from the assistant that was
causing misbehavior. Going forward, our intent is to always let you
observe and edit text before we send it.
---------
Co-authored-by: Marshall <marshall@zed.dev>
This PR refines the UX of editing a previous message, including the
following:
- Focus the prev message body editor on double-click
- Restore previous body text on cancel
- Cancel pending completion upon submission of previous message
- Drive-by: Remove min height on composer editor
Release Notes:
- N/A
---------
Co-authored-by: Marshall Bowers <marshall@zed.dev>
Nathan Sobo