feat: rustanka-language-server support#74
Draft
iainlane wants to merge 28 commits intografana:mainfrom
Draft
Conversation
Debounce file watcher updates and cancel in-flight eval requests. Track request ids so stale responses cannot overwrite newer output.
Create session-scoped temp file metadata and clean up both file and directory when the eval result document is closed.
Use async fs/process APIs for eval temp output and installer checks. This avoids blocking the VS Code extension host event loop.
Add redirect/time limits, verify sha256 assets when available, and install binaries with 0755 permissions.
Make the client nullable, add stopClient(), and harden restart/eval commands when the language server is not running.
Make resolvePathsWithTanka a string-only enum and resolve relative jpath entries against all workspace roots.
Split eval flow, importer command, LSP requests, settings mapping, and client lifecycle out of extension.ts.
Apply consistent whitespace and wrapping after the module split.
Enable @typescript-eslint/no-unused-vars and annotate intentional unused callback parameters with underscore prefixes.
Introduce a dedicated debug configuration provider and fill in type, request, name, program, and jpaths defaults.
Return the VS Code-provided executable descriptor when available, falling back to the bundled jsonnet debugger adapter.
Extract pure installer parsing/asset-selection helpers and cover version parsing, asset filtering, checksum matching, and hash parsing.
Add command-level usage details and explicit error behavior notes for active-editor, server-state, eval, and importer outcomes.
Use a TextDocumentContentProvider-backed jsonnet-eval scheme for live eval output updates and lifecycle cleanup.
Resolve debug jpath defaults from folder-scoped Jsonnet settings and normalize relative entries against the workspace root.
Replace the custom extension-host runner with @vscode/test-cli and a .vscode-test.js config. Add scenario-based integration tests under src/test/suite plus a test-scenarios/basic workspace that includes Jsonnet and Tanka trees, fake LSP/debugger binaries, and deterministic eval and diagnostics behavior. Also update docs/scripts, ignore local .vscode-test artifacts, and type githubApiRequest's encoding as BufferEncoding for newer @types/node.
Teach integration tests to use a real LSP binary when JSONNET_LSP_SERVER is set, while keeping fake-binary behavior as the default. Split .vscode-test.js into labeled fake/real configs and update npm scripts to run each label explicitly. The real label enforces JSONNET_TEST_REAL_LSP so missing JSONNET_LSP_SERVER fails fast. Keep language server and debugger auto-updates disabled in the harness for deterministic, non-interactive runs, and document the new real-mode workflow in README.
Add *.expected.json sidecars for test scenarios and make fake-jrsonnet-lsp load eval, diagnostics, and importer responses from those files instead of hardcoded branches. Update integration tests to read the same expected files, so fake-mode assertions are data-driven by scenario fixtures. For real LSP mode, keep eval assertions generic and skip diagnostics checks. Document the expected-file mechanism in README.
Replace element-by-element diagnostics assertions with deep structural equality against each scenario's expected diagnostics array. This removes first-element indexing from diagnostics checks and keeps the suite strict: real-LSP diagnostics now fail if expected diagnostics are not published.
Centralize only binary-path and auto-update test settings in the harness, while each test suite now sets its own language-server behavior flags via setup hooks. Drop manual `jsonnet.restartLanguageServer` calls from test setup. The suite passes in fake mode without explicit restarts, avoiding startup-race failures while still applying per-test configuration through workspace settings updates. Document that tests configure settings per case to avoid bleed-over.
Replace fake-only diagnostics fixtures with scenarios that map to real server diagnostics. Use `jsonnet/syntax_error.jsonnet` for parser error diagnostics and `jsonnet/unused_variable.jsonnet` for lint warning diagnostics. Update the diagnostics integration tests to target those files and remove the old deprecated-field scenario fixtures.
Move the integration workspace root to `test-scenarios` and split the former `basic` fixture tree into dedicated `commands`, `eval`, `diagnostics`, and `importers` scenario directories. Update integration tests to open files from their scenario directories so fixtures are explicit per suite. Adjust fake-jrsonnet-lsp scenario root discovery to detect the nearest workspace that contains `.tools/fake-jrsonnet-lsp.js`, which supports the new multi-scenario layout. Add importer fixtures that model real transitive imports and update the expected importer paths accordingly.
Refactor eval integration coverage into table-driven cases for JSON and YAML file evaluation, and extract shared jsonnet-eval output reading into a helper. Refactor diagnostics integration coverage into table-driven cases and remove length-based gating. Diagnostics now wait until the full current array is deep-strict-equal to the expected scenario output.
Add scenario-backed integration coverage for hover, navigation, rename, references, diagnostics matrix cases, formatting, completion, signature help, inlay hints, eval runtime errors, and Tanka mode detection. Refactor suite utilities into reusable modules, add real-LSP race coverage, and align fake-server expected file loading with the new scenario layout. Enforce deterministic harness defaults for binary paths and auto-update settings while keeping per-test configuration in each suite.
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
1 participant
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.
Introduce support for the highly experimental and buggy Rustanka language server, which is a Rust implementation of a Tanka/Jsonnet language server.
All of the server's settings are exposed.
There are currently some failing tests. I tried to keep these to cases where the server is (arguably) returning the wrong results.
How to run tests
From repo root:
To run integration tests against a real Rustanka binary:
JSONNET_LSP_SERVER=/absolute/path/to/jrsonnet-lsp npm run test:integration:realThe real run requires
JSONNET_LSP_SERVERand uses the same scenario suite, so fake and real modes are directly comparable.