ci / build (push) Canceled after 0s
In managed mode the extension switches models by rewriting the daemon config's defaultModel and restarting it, but loadState() preferred the persisted model, so the daemon came back on the old one every time. State now records the default it was saved under; a different default wins. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01J39imRDT8L7xYbWzsVWd8A
4.8 KiB
4.8 KiB
Changelog
Extension versions are released as vscode-v<version> on the Releases page. Models are versioned by their name in models/manifest.json.
vscode 0.6.3
- Fix: model choice didn't stick in managed mode. The daemon's saved state (last model used) overrode both the
model picked in the menu and a new default from a release, so it kept restarting on the old model (e.g.
stage-c). A changed default now wins; switching through the daemon API still persists as before.
vscode 0.6.2
- Next-edit is ~3–5× faster: the next-edit server uses n-gram speculative decoding (the rewritten region is mostly a copy of the current one, so drafted tokens are verified in batches). Same output, since decoding is greedy. Rewriting a 13-line region on a laptop CPU: ~2.1 s → ~0.7 s (~0.4 s when the prompt is cached).
- Next-edit timings in "Show Last Prompt" now split TS errors / prompt / generation time.
vscode 0.6.1
- New default model
stage-d, withstage-efor next-edit. Both download automatically (~530 MB each); the daemon runsstage-ein a second llama-server, only for next-edit.stage-dcloses blocks (}) and predicts the next line far better (held-out exact match: block close 0.27 → 0.79, next line 0.01 → 0.18), and sees TypeScript errors near the cursor. Results in models/README.md. - Fix: a "no edit" answer from a next-edit model could be applied as deleting the suggested region. llama-server drops the model's no-edit token, so it arrived as empty text. Next-edit models are only enabled from this version on.
- Manifest:
nextEditModel(companion model) andnextEditFormat(replacesnextEdit, which older versions ignore from now on).
vscode 0.6.0
- Next-edit suggestions (with models trained for it, e.g. the upcoming
stage-e): after you pause typing, the model may suggest an edit around the cursor, such as removing a leftover)or updating a renamed variable's uses. Single-line changes show deleted text struck through and inserted text as ghost text; multi-line changes are highlighted with the diff on hover. Tab applies, Esc dismisses. Setting:nextEdit.enabled. - Token healing: completions that start mid-word (
this.connections.del▮) now finish the word instead of skipping it. - Suggest on cursor move (
triggerOnCursorMove, off by default): with next-edit models it predicts what you came to change; otherwise it only fires on empty lines, line ends and lines with errors. - Pre-warm on focus (
prewarmOnFocus, off by default): fills the model cache when you switch files, so the first keystroke is fast. - Show Last Prompt command: the exact prompt the model saw, its raw output and timings.
- Update check against the Releases page (daily;
checkForUpdates), with one-click install. - Edit recorder also records VS Code's diagnostics (TS, Vue, ESLint) for recorded files.
- Daemon: model profiles carry the prompt format they were trained with (
contextFormat,errors,nextEdit), so newer models get TS errors and namespace-import symbols in their context.
vscode 0.5.0
- New repository layout (
models/,extensions/vscode/,daemon/,shared/). The model manifest moved tomodels/manifest.json, and the defaultlocalAutocomplete.modelManifestUrlfollows it. - Publisher is now
harvmaster(extension idharvmaster.local-autocomplete). Uninstall the oldlocal.local-autocompletefirst. The model is downloaded once more, into the new extension's storage. - MIT license, repository links, versioned VSIX file names.
vscode 0.4.0
- Managed mode (default): downloads the model (from
models/manifest.json) and the pinned llama.cpp build on first use, SHA-256 verified and resumable, then runs the bundled daemon on VS Code's own Node runtime. Shared by all windows; exits when idle; replaced automatically after updates. - Custom models via
localAutocomplete.modelPath+modelProfile; custom llama-server viallamaServerPath. - External mode (
daemon.mode: "external") keeps the previous behaviour.
vscode 0.3.0
- Recording-only mode (
enableCompletions: false+recordingDir), cursor-move and file-switch events, events tagged with the editor app (Cursor notice).
vscode 0.2.0
symbolscontext mode (TypeScript language service, same code as training data generation).- Missing-
this.repair:getU→this.getUser(id)when the name only exists on the class. - Opt-in edit recorder (
recordEdits). - Fix: a function's closing
}was being trimmed when the next line also closed a block.
vscode 0.1.0
- First version: inline completions via the acd daemon, context-mode toggle in the status bar, cancellation, exclusion globs.