Release v0.5.0#155
Merged
Merged
Conversation
These 82 tests each parsed a snippet and asserted only `ast` defined + `errors` empty — no AST-structure or codegen assertions. The constructs they touch (arithmetic/comparison/logical operators, control flow, literals, variable/type declarations, multi-POU) are all exercised far more rigorously by the codegen-*.test.ts and cpp-compile/st-validation suites, which parse *and* generate *and* assert real C++ output. A parse-only "no errors" check is strictly weaker and adds no coverage those provide. Removed in full: Function/FB Parsing, Expression Parsing, Chained Expression, Literal Types, Direct Address, Type Declaration, Additional Expression Edge Cases, Multiple POU, Variable Declaration, Control Flow Parsing, Edge Case. Trimmed: Parse Function Tests and Complex Program Tests keep only their non-shallow cases. Deliberately KEPT (not shallow): the negative test (`returns errors for invalid syntax`), the empty-input edge test, and the compile()-based VAR_GLOBAL/VAR_EXTERNAL success test. Supersedes #79, which missed the two function-parsing sections (they sit before its "keep first 536 lines" cutoff) and over-removed the compile()-based Complex Program test. 136 -> 54 tests; full suite green (1917 passed), typecheck clean. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
…tests test: remove shallow parse-only smoke tests from compile.test.ts (#72)
Two of the three issues in #75 are still valid; the third was already fixed. 1. oscat-gpp-compile.test.ts — the "compiles and runs full OSCAT" test silently `return`ed when no transpiled output existed, so it could pass without testing anything. The suite is already gated by `describe.skipIf(!hasGpp || !oscatStlibAvailable)`, so a missing artifact means the preceding transpile test failed — a real failure. Replaced the silent return with `expect.fail(...)`, and removed a dead `return` after an existing `expect.fail`. 2. located-variables.test.ts — "error on located variable in FB" asserted `message.includes('FUNCTION_BLOCK')`, a generic token any unrelated FB error would satisfy (vacuous match). Anchored it to the distinctive rule text "Located variables can only be declared …" so the test fails if a different error fires or the located-var check regresses. (CompileError has no populated `code` field — introducing an error-code taxonomy across the analyzer would be a much larger change, out of scope here.) 3. codesys-import.test.ts "V3 vs V2.3 counts" — already uses `toBeGreaterThan(...)` tolerance checks, not the brittle exact `toBe(v23fn)` equality the issue described. No change needed. Supersedes the stale, conflicting PR #83. Affected files pass (76 tests). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
…tions test: fix fragile/vacuous assertions (#75)
Two test files lived in the wrong layer: - tests/semantic/references.test.ts — only calls parse() (17 calls, no semantic analysis), so it's a parser-level test → moved to tests/frontend/ - tests/frontend/ast-utils.test.ts — runs the full analysis pipeline via analyze() to exercise findEnclosingPOU → moved to tests/integration/ Pure git renames (100% content unchanged). All test dirs are at depth 2, so the `../../src/index.js` imports stay valid; vitest discovery is glob (tests/**/*.test.ts), so location doesn't affect collection. Full suite unchanged: 1917 passed / 7 skipped. Supersedes PR #82, which moved only references.test.ts (1 of the 2 files) and was based on a stale March commit. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
test: move misplaced test files to correct directories (#74)
Hovering a built-in IEC 61131-3 type name (INT, TIME, LREAL, …) now shows a markdown tooltip with the full name, category, value range, bit size, and literal syntax where applicable. - New server/src/iec-type-docs.ts: IEC_TYPE_DOCS map (26 elementary types) + getIecTypeDoc(word). Sizes reflect how STruC++ actually lowers each type per src/runtime/include/iec_types.hpp. - hover.ts getHover: resolve a built-in type name from the source word at the cursor (type names are reserved keywords, so they never collide with user symbols and often have no AST symbol node). The server already advertises hoverProvider and routes onHover → getHover, so no wiring change was needed. Implemented fresh against the current hover architecture rather than reviving the closed PR #56, which targeted a stale feature branch, added a separate hover-provider.ts that doesn't fit today's getHover/handler flow, and — critically — documented TIME/DATE/TOD/DT as 32-bit, whereas STruC++ represents all temporal types as signed int64 (TIME/LTIME/TOD/DT in ns, DATE in days). The new docs are corrected accordingly. Tests: unit coverage for getIecTypeDoc (all 26 types, case-insensitivity, null cases, int64 temporal regression guard) + getHover integration (hovering INT/TIME shows docs; hovering a variable name does not). Full extension suite green (252 tests); tsc build clean. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
feat(vscode): hover docs for built-in IEC data types (#55)
The Phase 3.2/3.3 "validation example" tests (derived from old phase docs) re-tested constructs already covered by the dedicated codegen tests in the same files, with equal-or-weaker assertions. Removed 8: codegen-control-flow.test.ts (5): nested IF with ELSIF, FOR loop with EXIT, WHILE loop, REPEAT-UNTIL, CASE with ranges. Each duplicates a dedicated Phase 3.2 test (e.g. "EXIT as break" is actually stronger — it also asserts the exit label; REPEAT-UNTIL was byte-identical). codegen-composite.test.ts (3): the "Phase 3.3: Validation Examples" section (array access Test 1, structure access Test 2, array of structures Test 3) — kitchen-sink re-tests of the dedicated Array Element Access / Structure Member Access / Combined Access tests (Test 3 was identical to "array of struct access in loop"). Coverage preserved: the only non-identical case — that a CASE range (90..100) expands to *intermediate* labels — is already covered by the "CASE with range expansion" test (case 4..6 → case 5); for extra rigor I also added a `case 91:` assertion to "CASE with mixed labels and ranges" when removing the validation copy. Full suite: 1917 → 1909 (−8), 7 skipped; no coverage lost. Supersedes the stale PR #81, which conflicts against current development, re-adds an already-present compileST helper, and blind-deleted the CASE test without preserving the intermediate-expansion assertion. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
…odegen-tests test: remove duplicate codegen test sections (#73)
…ke (#71) Behavior tests were importing the compiled dist/ output. Because coverage is collected on src/** (dist/ excluded, not remapped), those tests contributed ZERO to src/ coverage — e.g. codegen-composite alone now attributes ~46% to src/backend/codegen.ts where before it counted 0%. The "fails without a build" rationale in the issue is moot (the vitest globalSetup runs tsc first), but the coverage-attribution problem is real. - Convert all 14 dist-importing test files to ../…/src/… (vite resolves the .js specifiers to .ts, as the rest of the suite already relies on). The grep-based pass found 13; the new guard caught a 14th at the tests/ root (diagnostic-pou-context.test.ts, one ../ deep). - Add tests/no-dist-imports.test.ts: a guard test that fails if any test imports dist/. Implemented as a test rather than ESLint because `npm run lint` only covers src/ and tests are excluded from tsconfig, so type-aware ESLint can't lint them without a separate project. - Add tests/integration/package-smoke.test.ts: the one deliberate dist/ consumer — imports the shipped dist/index.js and asserts the public API surface + a trivial end-to-end compile, validating tsc emit/exports/ESM resolution that src/ tests (vite-transpiled) don't. Allow-listed by the guard. Full suite: 74 files, 1913 passed / 7 skipped. Supersedes the stale PR #77 (8 of 13 files, conflicts on current dev). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
test: import src/ not dist/ in tests + dist guard & package smoke (#71)
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
chore(release): v0.5.0
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.
Promote development to main for the v0.5.0 release.
Since v0.4.22:
src/(so they count towardsrc/coverage) + a guard test banningdist/imports + a deliberatedist/package-smoke test.compile.test.ts.expect.fail; located-vars message anchoring).Tagging
v0.5.0after merge kicks the release build.🤖 Generated with Claude Code