Repository navigation
Fix false prompt-injection alarm on PR Tooling Safety Check bypass labels - #20130
Merged
Merged
Conversation
…bels The threat-detection job is a separate LLM that only sees the workflow description plus the agent's output, not the process steps. When the agent correctly applies AI-Tooling-Check-Bypassed to a non-fork PR, the detector misreads the bypass label as the agent being manipulated into skipping its scan and raises a false prompt-injection alarm, aborting the run's label and memory outputs. Give the detector context via threat-detection.prompt. Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com>
github-actions
Bot
added
the
AI-Tooling-Check-Bypassed
…ck-false-injection
pull_request_target grants a read-only GITHUB_TOKEN whenever the pull request modifies files under .github/workflows/ (a GitHub security restriction). The release-notes summary comment then fails with HTTP 403 and fails the whole check, even though the release-notes gate itself passed. Wrap the comment write in a try/catch that skips on 403 with a notice, so posting the comment stays best-effort and workflow-modifying PRs are no longer blocked by this check. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
The release-notes comment failure is a non-required check that fails for every PR modifying files under .github/workflows/: pull_request_target grants a read-only GITHUB_TOKEN in that case, so the comment write returns 403. This cannot be fixed from a PR branch because pull_request_target runs the base-branch (main) workflow, and it is already addressed by the dedicated PR #20136. Keep this PR focused on the labelops threat-detection fix. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
…ck-false-injection
…t 403 Rewriting the release-notes check in #20081 reduced the token to pull-requests: read, but posting a NEW PR comment (issues.createComment POST) requires pull-requests: write. Brand-new PRs with no existing bot comment (like #20130 and dependency PRs) therefore fail with 403 'Resource not accessible by integration', while PRs with a pre-existing comment pass via the updateComment PATCH path (allowed by issues: write). Restore pull-requests: write (the fork-head checkout removed in #20081 stays removed, so no untrusted code runs and the comment body is passed via env var) and wrap the comment write in try/catch so the best-effort informational comment can never fail the release-notes gate. Supersedes the abandoned PR #20136. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
…ck-false-injection
Member
Author
|
Note on the red
This branch already contains the fix (restore
|
…ck-false-injection # Conflicts: # .github/workflows/check_release_notes.yml
Contributor
✅ No release notes required |
T-Gro
commented
Aug 5, 2026
T-Gro
left a comment
T-Gro
left a comment
Member
Author
There was a problem hiding this comment.
🤖 This review was generated by AI (@expert-reviewer agent). Findings may contain inaccuracies — please verify independently.
Reviewed the threat-detection hint change and its compiled .lock.yml. The generated lock file is consistent with the .md source (description + CUSTOM_PROMPT env match, frontmatter_hash regenerated), and CUSTOM_PROMPT is consumed by the standard setup_threat_detection.cjs step. One security-relevant scoping concern noted inline.
Address review feedback: the detector hint previously told the threat detector that applying AI-Tooling-Check-Bypassed to any PR is benign. Bypass-labeling is only the designed path for non-fork PRs; a fork PR carrying a bypass label instead of a diff scan is exactly the injection outcome this detector should catch. Scope the reassurance to the non-fork bypass / fork scanned-clean path and explicitly flag a bypass label on a fork PR as a deviation worth flagging. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
…ck-false-injection
T-Gro
enabled auto-merge (squash)
August 6, 2026 12:52
abonie
approved these changes
Aug 6, 2026
github-project-automation
Bot
moved this from New
to In Progress
in F# Compiler and Tooling
github-project-automation
Bot
moved this from In Progress
to Done
in F# Compiler and Tooling
bartelink
pushed a commit
to bartelink/fsharp
that referenced
this pull request
Aug 6, 2026
…bels (dotnet#20130) * Fix false prompt-injection alarm on PR Tooling Safety Check bypass labels The threat-detection job is a separate LLM that only sees the workflow description plus the agent's output, not the process steps. When the agent correctly applies AI-Tooling-Check-Bypassed to a non-fork PR, the detector misreads the bypass label as the agent being manipulated into skipping its scan and raises a false prompt-injection alarm, aborting the run's label and memory outputs. Give the detector context via threat-detection.prompt.
This was referenced Aug 7, 2026
T-Gro
added a commit
that referenced
this pull request
Oct 16, 2026
* Avoid leaking a MeterListener per Cache in DEBUG builds (#19995) * Avoid leaking a MeterListener per Cache in DEBUG builds In DEBUG builds, every Cache instance created a CacheMetrics.CacheMetricsListener, which starts a System.Diagnostics.Metrics.MeterListener registered in the process-global metrics registry. These were never disposed, so they accumulated for the lifetime of the process. Because every cache hit/miss/add publishes a measurement to all registered listeners, the per-operation cost grew linearly with the number of leaked listeners, so workloads that create many caches (for example repeated ParseAndCheckProject / per-file checks) slowed down steadily. Track the per-cache totals used by DebugDisplay directly, incrementing a small Stats object alongside the existing global Meter counters, instead of via a per-cache MeterListener. No listener is created, so nothing leaks, and DebugDisplay still works. The now-unused CacheMetrics.Hit/Miss/Add/Update/Eviction/EvictionFail helpers are replaced by a single recordMetric helper. * Address review: drop per-cache CacheMetricsListener and cacheId tag - Remove the CacheMetrics.CacheMetricsListener type. Its only per-cache use was the #if DEBUG debugListener each Cache created and never disposed, which was the leak this PR set out to fix. (majocha) - Drop the per-instance cacheId tag (and nextCacheId). Measurements now carry only the cache name, shrinking the payload published to any connected exporter and removing the per-instance filtering that was cacheId's only purpose. (majocha) - DebugDisplay and the cache tests read the existing name-aggregated stats via CacheMetrics.getTotalsByName / getRatioByName, populated by the single process-wide ListenToAll listener. No per-cache listener is created and no per-operation cost is added in any configuration, so there is no DEBUG-only overhead left to gate behind a separate directive. (T-Gro) - Overload-cache tests enable ListenToAll and snapshot totals before/after to stay scoped to their own compilation; FSharpChecker .CreateOverloadCacheMetricsListener is removed. * Update public SurfaceArea baseline after removing CacheMetricsListener CacheMetricsListener was a public type, so dropping it changes the recorded public surface. Remove its 10 entries from FSharp.Compiler.Service.SurfaceArea.netstandard20.bsl; the SurfaceArea test now passes. Also note the single-listener assumption the cache metric tests rely on. * Apply fantomas formatting to Caches.fs * Document why OverloadCacheTests is not parallelizable (global cache metrics state) --------- Co-authored-by: Tomas Grosup <Tomas.Grosup@gmail.com> (cherry picked from commit 7192705) * Add ResetCompilerGeneratedNameState to compiler-generated name generators (#20017) Compiler-generated occurrence names (name@line-N) are allocated from process-wide counters on CompilerGlobalState that accumulate across compilations. When a warm checker re-emits the same project in-process, an unchanged closure therefore gets a different occurrence suffix than the previous emit, so consumers that align generated names across compilations (Edit-and-Continue delta emission, #19941) cannot match them. Add an internal ResetCompilerGeneratedNameState to NiceNameGenerator (clears the per-(name, file) occurrence counters), StableNiceNameGenerator (clears the cached stable names and the inner counters), and an aggregate on CompilerGlobalState that resets all three generators, restoring the fresh-process name layout. Callers must ensure no compilation is concurrently generating names. No in-tree caller yet; the consumer is the hot reload emit path in #19941. Covered by unit tests proving drift without reset, exact replay after reset, and that the stable-name cache itself is cleared. (cherry picked from commit 60d315a) * Localized file check-in by OneLocBuild Task: Build definition ID 499: Build ID 3013177 (#20023) Co-authored-by: Copilot <copilot@github.com> (cherry picked from commit 9fc230a) * Tests/source context: support multiple carets (#20077) (cherry picked from commit 69fca7f) * Move VS language-service logic tests to FSharp.Compiler.Service.Tests (#20033) * Move VS language-service logic tests to FSharp.Compiler.Service.Tests Port completion, quick info, parameter info, go-to-definition, and diagnostics coverage from the Windows-only VS Salsa suite to the cross-platform FSharp.Compiler.Service.Tests. The legacy suite keeps only the tests that genuinely exercise Visual Studio integration. (cherry picked from commit 2f07589) * Add Roslyn-format EnC CustomDebugInformation codec and portable PDB method CDI emission (#20018) * Add Roslyn-format EnC CustomDebugInformation codec and portable PDB method CDI emission Adds an internal AbstractIL module implementing, byte for byte, the three Portable PDB CustomDebugInformation blob formats Roslyn persists per method for Edit and Continue (EnC Local Slot Map, EnC Lambda and Closure Map, EnC State Machine State Map), with serializers, deserializers, a portable PDB read-back helper, and an occurrence-key packing helper for deterministic syntax-offset slots. Plumbs an optional methodCustomDebugInfoRows side channel through the IL binary writer options into the portable PDB generator so a compilation can attach CDI rows to named methods. Names that do not identify exactly one method row are dropped. All existing writer call sites pass an empty map, so emitted PDBs are byte-identical to before. No in-tree caller populates the map yet; the consumer is the F# hot reload work in #19941, following the same pattern as #20017 (land isolated, test-covered infrastructure first, wire the feature later). Tests: blob round-trips, Roslyn golden-byte encodings, cross-validation against CDI blobs emitted by a real Roslyn compilation, fail-closed occurrence-key packing (including an int32-overflow regression where a wrapped negative key previously escaped the bound check), and end-to-end synthetic PDB emission proving correct MethodDef parenting, zero rows for an empty map, and no rows for absent or ambiguous names. (cherry picked from commit 24d731a) * Add stable synthesized-name replay infrastructure for hot reload (#20024) * Extract stable synthesized-name replay layer Add internal generated-name normalization and synthesized-name map replay support as a standalone slice. The new map state is side-channel based, all new compiler modules remain internal, and CompilerGlobalState preserves the existing no-map counter path while checking an accessor captured once per compiler state. Route existing IlxGen generated-name allocations through inert helper wrappers, add pure name-map and normalizer tests, add a normal compilation determinism guard over emitted generated names, and document the extracted seams in P5_REPORT.md. Verification: built FSharp.Compiler.Service, FSharp.Compiler.Service.Tests, FSharp.Compiler.ComponentTests, and FSharpSuite.Tests in Release; ran the migrated service test classes, the component determinism class, FSharpSuite DeterministicTests, and the FCS SurfaceArea class successfully. * Fix generated-name scope test in stable names slice * Validate hot reload generated names before classification * Format hot reload compiler sources Verified with the repository-wide Fantomas check. * Retry CI after Linux runner memory exhaustion * Make synthesized name snapshots deterministic (cherry picked from commit 2e838bb) * Fix #19457: lift CE constructs from plain let RHS in computation expressions (#19868) (cherry picked from commit 4a2749e) * Fix attribute resolution in recursive module/namespace scopes (#19744) (cherry picked from commit 1d8dc39) * Correct StructLayout size emission for data-less struct unions (#19759) (cherry picked from commit 1dc395a) * Report FS3888 for generic attribute type abbreviations instead of FS0193 (#19915) (cherry picked from commit 8c0e444) * Support NotNullIfNotNullAttribute (#19977) (cherry picked from commit 0c94e4e) * Compiled ToStrings under -reflectionfree for DUs and Records (#19976) * Add a compiler intrinsic for the 'string' operator Adds string_operator_info / mkCallStringOperator so generated code can call Operators.string. These lines are duplicated by the interpolated-string PR (#19971); kept identical there so a future merge resolves cleanly. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * Generate a match-based ToString for unions under --reflectionfree Under --reflectionfree the union ToString previously emitted nothing, so DUs fell back to Object.ToString() (the namespace-qualified type name). Instead generate a match over the cases that builds "CaseName(f0, f1, ...)" using the 'string' operator on each field, via a TypedTree expression fed to CodeGenMethodForExpr. This recurses naturally into nested unions and is reflection-free. The default (sprintf "%+A") path is unchanged. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * Extract mkStringConcat helper for arity-dispatched String.Concat The "concatenate a list of string exprs, picking the cheapest String.Concat overload by arity" pattern was duplicated in CheckExpressions (interpolation lowering) and the optimizer, and our new union ToString used the array overload unconditionally. Extract mkStringConcat into TypedTreeOps.ExprOps and route all three through it. This also lets single-field union cases emit Concat3 instead of allocating a string[] (IlxGen runs after the optimizer, so nothing else would collapse that array form). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * Fix generated union ToString for generic unions The match-based ToString body is a TypedTree expression codegen'd via CodeGenMethodForExpr, but it was built with `eenv`, which lacks the tycon's type parameters. For generic unions this produced wrong IL: the wrong case branch (always the null-as-true-value case) or a NullReferenceException for single-case unions. Use `eenvinner` (the per-tycon environment) so the generic method body resolves its type parameters. The old sprintf path was unaffected because it emits raw IL off the pre-built ilThisTy. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * Render union ToString fields like option (null -> "null") To make a generated union ToString consistent with how option/list format their contents (LanguagePrimitives.anyToStringShowingNull), format each field as: if (box field) is non-null then 'string field' else "null". Previously a null field rendered as "" (the 'string' operator's null behaviour). Generated inline rather than calling anyToStringShowingNull, which is internal to FSharp.Core and so not callable from user-compiled code. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * Tidy reflection-free union ToString tests Normalize union declarations to a leading '|', use System.Console.WriteLine instead of printfn (the printf machinery is what these changes move away from), and make the null-field test compare the union's rendering directly against option's rather than asserting a fixed string. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * Add reflection-free ToString to Result and Choice Result and Choice had no ToString override, so they fell back to the compiler-generated sprintf "%+A" one, which uses reflection. Give them hand-written overrides mirroring option/list (String.Concat + anyToStringShowingNull), e.g. Ok 5 -> "Ok(5)", Choice1Of2 7 -> "Choice1Of2(7)". This is reflection-free / AOT-friendly and consistent with option's "Some(x)" rendering. Note: this changes the observable ToString of Result/Choice from the "%A"-style "Ok 5" to "Ok(5)". Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * Generate a single-line ToString for records under --reflectionfree Records previously fell back to Object.ToString() (the namespace-qualified type name) under --reflectionfree. Generate "{ F1 = v1; F2 = v2 }" on a single line (no line breaks, unlike sprintf "%+A"), with fields formatted like union fields (null -> "null", otherwise via 'string'). Factor the shared field formatter and ToString-method emission out of the union path. The default (sprintf "%+A") path is unchanged. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * Update FSharp.Core surface-area baselines for Result/Choice ToString Result and Choice`2..7 now declare an explicit ToString() override, so they appear in the public surface area. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * Add release notes Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * Generate a single-line ToString for anonymous records under --reflectionfree Drive anonymous-record ToString through the synthetic record tycon (already built for equality/comparison) rather than sprintf "%A", so under --reflectionfree it renders "{| Name = value; ... |}" on a single line. GenRecordToStringMethod now takes open/close brace strings ("{ "/" }" for records, "{| "/" |}" for anonymous records). The default (non-reflection-free) codegen path is unchanged and still falls back to sprintf "%+A". Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * Test that a hand-written ToString override is kept under --reflectionfree Addresses review feedback: generation is gated on `not (HasMember "ToString")`, so a user-defined ToString on a union or record wins over the generated one. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * Rename ToString generators for clarity Addresses review feedback: distinguish the reflective sprintf path from the structural one. GenPrintingMethod -> GenSprintfPrintingMethod (the sprintf "%+A" ToString/get_Message), GenToStringMethodFromExpr -> EmitToStringMethodDef. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * Restore tabular layout for string_operator_info in TcGlobals Addresses review feedback: keep the column-aligned layout of the surrounding intrinsic table. Also makes these two lines byte-identical to the same intrinsic added by #19971, so a future merge resolves cleanly. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * Add reflection-free ToString tests for field shapes, structs, anon records and recursion Covers DU field shapes (multiple fields vs a single tuple field), explicit vs unnamed field names rendering identically, struct unions/records, anonymous and struct anonymous records, and finite recursive/nesting types. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * Add EmittedIL tests for reflection-free record and union ToString Locks in the IL emitted under --reflectionfree: each field is boxed and rendered through Operators.ToString with a null guard, and the parts are joined with String.Concat (array form for the record, 3-arg form for the single-field union case). Nullary union cases return the bare case name. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * Generate reflection-free ToString in the augmentation phase The structural ToString for --reflectionfree records and unions was built in IlxGen, after the optimizer, so its per-field 'string' operator calls were never inlined: each value-type field was boxed and rendered through the generic Operators.ToString, behind a null guard that is dead for a value type. Move the generation into the type-augmentation phase (alongside Equals/GetHashCode/CompareTo) so the body flows through the optimizer. The 'string' operator is now specialised - a value-type field renders via a direct, allocation-free invariant-culture ToString with no boxing and no null guard (reference fields keep the guard so null still renders as "null"). The shared body builders live in AugmentTypeDefinitions; anonymous record types are synthesized too late for augmentation, so they keep generating in IlxGen but reuse the same builder. Output is unchanged; the EmittedIL baselines are updated to the leaner IL. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * Guard generated reflection-free ToString against deep-recursion overflow The augmentation-generated structural ToString recurses into fields, so a deeply nested value can exhaust the stack with an uncatchable StackOverflowException. Emit RuntimeHelpers.EnsureSufficientExecutionStack() at method entry (as C# records do in PrintMembers) so it throws a catchable InsufficientExecutionStackException instead, when the runtime provides the method. The guard is skipped for types whose every field is a flat primitive (integer/float/decimal/string/char/bool/unit/enum), which cannot recurse. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * Test the reflection-free ToString deep-recursion guard A 1,000,000-deep value's generated ToString throws a catchable InsufficientExecutionStackException rather than hard-crashing the process. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * revert ToString additions to fsharp.core types * Remove stale FSharp.Core release note for the reverted Result/Choice ToString Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * Fix code formatting in IlxGen.fs (dotnet fantomas) Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * don't use quoted name * int version of reflectionfree-printing doc * doc tweaks * Link release note to the printing doc and cover anonymous records Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * test backticks * Share the ToString recursion guard with anonymous records The guard lived in MakeBindingsForToStringAugmentation, which anonymous records bypass: they are synthesized too late for type augmentation and reach mkRecdToString from IlxGen instead. Deep nesting overflowed the stack rather than raising InsufficientExecutionStackException. Move it into mkToStringRecursionGuard, applied inside mkRecdToString and mkUnionToString, so every caller of the body builders gets it. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * Add EmittedIL baselines for struct and anonymous record ToString Struct records and unions read fields off the this pointer and switch on the tag, and the anonymous record path is generated separately in IlxGen, so each gets its own baseline. The anonymous baseline omits the field reads: they name the anonymous type, whose mangled name is not stable across compilations. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * Fix empty anonymous record ToString rendering a doubled space The open/close braces carry inner spaces ("{| " and " |}"); with no fields they abut and render "{| |}". Trim the leading space when the field list is empty, matching %A's "{| |}". Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> * tidy comment --------- Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> (cherry picked from commit 17cb503) * Run ilverify via the tool manifest instead of a hard-coded cache path (#20101) (cherry picked from commit c00299f) * Record spreads (#18927) (cherry picked from commit f4b785f) * Implement interpolated strings via String.Concat (#19971) (cherry picked from commit d3403ca) * Secure release-note checks for fork pull requests (#20081) * Secure release-note checks for fork pull requests * Address release-note workflow review feedback (cherry picked from commit 647548b) * Update test project to net11 (#20104) * Update test project to net11 Internal CI was failing since the move to net11 because restoring this test project had to suddenly be done via network call to nuget.org * Update target framework and PDB path in tests (cherry picked from commit 536800c) * LexFilter: drop non-strict mode (#20106) (cherry picked from commit ea778bb) * Implement direct delegates (#19993) (cherry picked from commit d89529c) * Fix check_release_notes 403 by restoring pull-requests: write and making the comment non-fatal (#20198) check_release_notes runs via pull_request_target, so GitHub executes the workflow from the default branch (main). Creating the informational PR comment requires pull-requests: write, but #20081 reduced the token to read, turning the check red with HTTP 403 on any PR that had to create (not update) the comment - e.g. Maestro/darc PR #20133. Restore pull-requests: write so the comment posts, and guard the comment step with continue-on-error plus try/catch so posting can never fail the release-notes verdict. Supersedes #20200. (cherry picked from commit 92a3700) * Adopt ordered multi-caret markers in FCS tests (no behaviour change) (#20082) * Clean up goto-def tests: drop unused open System and redundant caret-anchor comments Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Copilot-Session: 9d20cd22-45dc-411f-9163-185fd6dd54d9 * Collapse copy-pasted completion sources with ordered {caretN} markers Uses SourceContext.extractOrderedMarkedSources (multi-caret) to replace whole-source [<Theory>] copies with one marked source in DotOff.ArraySliceNotation (3 copies) and CurriedArguments.Regression (5 copies). No behaviour change. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Copilot-Session: 9d20cd22-45dc-411f-9163-185fd6dd54d9 * Collapse copy-pasted completion sources in PatternMatching and Generics Bug312557_2 (4 source copies -> 1 {caretN}) and Bug69673_1.CtrlSpaceForThis (2-row Theory -> 1). No behaviour change. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Copilot-Session: 9d20cd22-45dc-411f-9163-185fd6dd54d9 * Collapse Symbols 'Nested copy-and-update' 8 copied sources into one {caretN} The 8 Facts shared an identical source copied per caret; now one {caret1..8} source + a (field-name, range) cases list. No behaviour change. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Copilot-Session: 9d20cd22-45dc-411f-9163-185fd6dd54d9 * BreakpointLocation: mark expected ranges with {selstart}/{selend} in source Replaces magic ((line,col),(line,col)) tuples with {selstart}/{selend} markers around the breakpoint span; the validation caret is inferred from {selend} (SourceContext), so no {caret} needed. Compares against context.SelectedRange. No behaviour change (all 6 tests green). Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Copilot-Session: 9d20cd22-45dc-411f-9163-185fd6dd54d9 --------- Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Copilot-Session: 9d20cd22-45dc-411f-9163-185fd6dd54d9 (cherry picked from commit 2038269) * Record spreads: off-by-default shadowing warnings (#20206) (cherry picked from commit 05af9c8) * Remove always-on PrintfBinaryFormat language feature flag (#20202) * Add RED tests for unconditional %B and rejected --disableLanguageFeature:PrintfBinaryFormat Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Remove always-on PrintfBinaryFormat language feature flag Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Correct langversion comment in disableLanguageFeature %B guard test Fixes an inaccurate comment (claimed minimum accepted --langversion is 8.0) flagged during expert review. %B is now unconditional across all langversions after removing the PrintfBinaryFormat feature flag. Comment-only change. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Make %B guard test accurate: assert compile+run at supported langversion The expert-review finding noted the %B guard test compiled under the default (latest) langversion where %B was already accepted, giving no regression value for the removed 6.0 gate. A sub-6.0 test is infeasible: the minimum supported langversion is 8.0 (lower versions error with FS3880), already above the old 6.0 gate, so removing the PrintfBinaryFormat flag is a pure no-op for every supported langversion. The guard test now clearly asserts the %B code path is intact (compiles and runs, producing 10011), with the 3881 rejection test covering removal of the feature name. Comment documents why no sub-6.0 test exists. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Add release notes for PrintfBinaryFormat flag removal Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Drop release notes and added guard tests: this is cleanup-only Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com> --------- Co-authored-by: Copilot <copilot@github.com> Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> (cherry picked from commit 5fa2950) * Stabilize preview language features into F# 11.0 (#20199) * Stabilize preview language features into F# 11.0 Move MethodOverloadsCache, ErrorOnMissingSignatureAttribute, DirectDelegateConstruction, AccessProtectedBaseFieldFromClosure and RecordSpreads from previewVersion to languageVersion110. FromEndSlicing stays in preview in its own block. Also relocate the misplaced ImplicitDIMCoverage into the F# 11.0 block and move the corresponding release notes from .Language/preview.md into .Language/11.0.md. Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com> * Fill PR number in release note Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com> --------- Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com> (cherry picked from commit 84eefce) * Implement `<inheritdoc>` XML documentation support for F# (#19188) (cherry picked from commit 93659f3) * Add support for `<include>` XML documentation tag (#19186) * Add support for <include> XML documentation tag (#19186) Implement support for expanding <include file="..." path="..."/> elements in XML doc comments when generating documentation files via --doc. (cherry picked from commit 5260b68) * Fix false prompt-injection alarm on PR Tooling Safety Check bypass labels (#20130) * Fix false prompt-injection alarm on PR Tooling Safety Check bypass labels The threat-detection job is a separate LLM that only sees the workflow description plus the agent's output, not the process steps. When the agent correctly applies AI-Tooling-Check-Bypassed to a non-fork PR, the detector misreads the bypass label as the agent being manipulated into skipping its scan and raises a false prompt-injection alarm, aborting the run's label and memory outputs. Give the detector context via threat-detection.prompt. (cherry picked from commit 2c245dd) * feat(Async): RunSynchronouslyImmediate (#19804) (cherry picked from commit 9357a4f) * Add ECMA-335 EnC metadata delta writer (#20019) * Add ECMA-335 EnC metadata delta writer Adds an internal, standalone ECMA-335 Edit-and-Continue metadata delta writer to AbstractIL: delta #- table stream and heap construction (DeltaMetadataTables, DeltaMetadataSerializer, DeltaTableLayout, DeltaIndexSizing), ECMA-335 II.24.2.6 coded-index encoding (DeltaMetadataEncoding), EncLog/EncMap emission, generation GUID chaining, user-string and standalone-signature token calculators (IlxDeltaStreams), and the coordinating writer (FSharpDeltaMetadataWriter) over a plain row-description input model (DeltaMetadataTypes, ILDeltaHandles, ILMetadataHeaps). The writer's inputs are row records (names, tokens, signatures, RVAs) plus heap offsets; it has no dependency on any semantic diffing or session machinery. It compiles with no in-tree consumer by design: the consumer is the F# hot reload work in #19941, following the same upstreaming pattern as #20017 and #20018 (land isolated, test-covered infrastructure first, wire the feature in a later PR). One line of ilwrite.fsi is touched to expose the pre-existing markerForUnicodeBytes so the delta writer reuses the exact string-marker logic of the full writer. No behavior change for any existing code path. Tests (130): coded-index encodings asserted against the production definitions and ECMA-335 II.24.2.6 order, System.Reflection.Metadata reader parity over emitted deltas, EncLog/EncMap correctness, stream layout, heap and index sizing, multi-generation heap-offset chaining asserted against computed expected values, standalone-signature rows asserted at baseline+1 from a real seeded baseline, and serializer failure paths. (cherry picked from commit e9fdc37) * Remove always-on StructActivePattern language feature flag (#20208) (cherry picked from commit 2def18d) * Remove always-on OpenTypeDeclaration language feature flag (#20209) (cherry picked from commit e1ad05c) * Allow closing '>' of multiline nested type arguments to align with the opener (#15171) (#20003) (cherry picked from commit 34a3053) * Record constructors (FS-1073) (#19974) (cherry picked from commit 7e2cbf1) * docs: update state-machine.md for changed workflow SHAs (#20234) Incremental update: - Fix LPSS safe-output label count (10 → 11) - Update source-shas footer Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com> Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> (cherry picked from commit 4a929f3) * Localized file check-in by OneLocBuild Task: Build definition ID 499: Build ID 3039021 (#20108) * Localized file check-in by OneLocBuild Task: Build definition ID 499: Build ID 3036509 * Localized file check-in by OneLocBuild Task: Build definition ID 499: Build ID 3036509 * Localized file check-in by OneLocBuild Task: Build definition ID 499: Build ID 3038596 * Localized file check-in by OneLocBuild Task: Build definition ID 499: Build ID 3038596 * Localized file check-in by OneLocBuild Task: Build definition ID 499: Build ID 3038733 * Localized file check-in by OneLocBuild Task: Build definition ID 499: Build ID 3038733 --------- Co-authored-by: Copilot <copilot@github.com> (cherry picked from commit ab2e066) * Enable MSBuild node reuse defaults in eng/Build.ps1 and eng/build.sh (#20220) Both entry points hard-disabled MSBuild node reuse for every build, overriding Arcade's default (node reuse on locally, off on CI). The overrides were added together in March 2019 (7a6448e, f4bd221) as a workaround for FSharp.Build.dll version conflicts when the proto compiler and the freshly built one are loaded into reused MSBuild nodes during F#'s self-bootstrap. This removes both overrides so Arcade's default applies, restoring node reuse for the inner loop and unblocking MSBuild Server. Whether the 2019 FSharp.Build.dll conflict still reproduces is validated by full CI, including the Windows EndToEndBuildTests job, which builds without -ci and therefore now exercises node reuse end to end. Fixes #20217 Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com> (cherry picked from commit 80b7e4e) * Fix spurious FS0410 for tuple patternInput bindings (#4161) (#19947) Skip accessibility checks on compiler-generated patternInput module bindings. These temps are module-init scaffolding whose visibility does not reflect the enclosing let-binding scope, which caused false FS0410 errors when tuple deconstruction referenced private types in the same module. Fixes #4161 (cherry picked from commit 853c500) * Ignore lines starting with `#:` (RFC-1337) (#20212) (cherry picked from commit 8111bff) * feat(Async+Task+ValueTask): consistent helper modules (#19844) (cherry picked from commit 4890294) * Add hot reload baseline reading and recorded EnC state (#20026) * Add Roslyn-format EnC CustomDebugInformation codec and portable PDB method CDI emission Adds an internal AbstractIL module implementing, byte for byte, the three Portable PDB CustomDebugInformation blob formats Roslyn persists per method for Edit and Continue (EnC Local Slot Map, EnC Lambda and Closure Map, EnC State Machine State Map), with serializers, deserializers, a portable PDB read-back helper, and an occurrence-key packing helper for deterministic syntax-offset slots. Plumbs an optional methodCustomDebugInfoRows side channel through the IL binary writer options into the portable PDB generator so a compilation can attach CDI rows to named methods. Names that do not identify exactly one method row are dropped. All existing writer call sites pass an empty map, so emitted PDBs are byte-identical to before. No in-tree caller populates the map yet; the consumer is the F# hot reload work in #19941, following the same pattern as #20017 (land isolated, test-covered infrastructure first, wire the feature later). Tests: blob round-trips, Roslyn golden-byte encodings, cross-validation against CDI blobs emitted by a real Roslyn compilation, fail-closed occurrence-key packing (including an int32-overflow regression where a wrapped negative key previously escaped the bound check), and end-to-end synthetic PDB emission proving correct MethodDef parenting, zero rows for an empty map, and no rows for absent or ambiguous names. (cherry picked from commit 51dd37e) * Optimizer: propagate NoDynamicInvocation (#20242) (cherry picked from commit 23fa8d5) * Support OverloadResolutionPriority and a most-concrete overload tiebreaker Two opt-in (`--langversion:preview`) additions to method overload resolution. - F# now honors `System.Runtime.CompilerServices.OverloadResolutionPriorityAttribute` (.NET 9): among the applicable candidates the highest-priority ones are considered first, matching C#. Priority is applied before betterness and scoped by the declaring type, so a library can steer F# callers to the overload it intends. - When overload resolution would otherwise be ambiguous, the more concrete candidate is now preferred — `Option<'T>` over `'T`, `Task<'T>` over `'T`, `Result<'T, string>` over `Result<'T, 'E>` — across constructors, static and instance members, extension members, and optional/ParamArray tails. Candidates that are incomparable (more concrete at different positions, or differing only by an SRTP constraint) stay ambiguous rather than being resolved arbitrarily. The tiebreaker is transparent: opt into `--warnon:3575` to see which concrete overload was selected and `--warnon:3576` for each generic overload it bypassed. RFC FS-1338 (OverloadResolutionPriority) and RFC FS-1340 (most-concrete tiebreaker). (cherry picked from commit 089835d) * Remove always-on UseBindingValueDiscard language feature flag (#20162) (#20218) Co-authored-by: Copilot <copilot@github.com> Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> (cherry picked from commit c2b38d7) * Remove always-on WildCardInForLoop language feature flag (#20221) * Remove always-on WildCardInForLoop language feature flag Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Add release note for WildCardInForLoop flag removal Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Strip release-notes changes (NO_RELEASE_NOTES) --------- Co-authored-by: Copilot <copilot@github.com> Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> (cherry picked from commit afe45bf) * Remove always-on AndBang language feature flag (#20223) * Collapse always-on AndBang enforcement gates in CheckComputationExpressions Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Remove always-on AndBang language feature flag and regenerate xlf Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Fix Fantomas formatting after AndBang gate removal The CheckCodeFormatting CI job failed because the collapsed AndBang gate left a two-line let binding that Fantomas collapses to a single line. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> --------- Co-authored-by: Copilot <copilot@github.com> Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> (cherry picked from commit 4990959) * Remove always-on ImplicitYield language feature flag (#20230) * Collapse ImplicitYield enforcement sites to always-enabled path Remove the dead 'SupportsFeature LanguageFeature.ImplicitYield' guards, whose condition is always true for langversion >= 8.0, across the four Checking/Expressions files. Each guard is collapsed to its enabled branch and the dead branch/binding deleted. Behaviour is unchanged; the flag definition in LanguageFeatures.fs is intentionally left for a later sprint. The now-unused 'cenv' parameter of YieldFree is renamed to '_cenv' (not removed): FSharp.Profiles.props enables --warnon:1182 with WarningsAsErrors, so an unused named binding is a hard build error, and removing the parameter would cascade through IsSimpleSemicolonSequenceElement, TryGetSimpleSemicolonSequenceOfComprehension and the SimpleSemicolonSequence active pattern (all out of scope). The '_' prefix preserves the exact signature and all positional call sites. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Remove ImplicitYield language feature flag definition The ImplicitYield feature shipped in F# 4.7 and is permanently enabled (minimum accepted --langversion is 8.0), making the flag dead configuration. Remove the LanguageFeature.ImplicitYield union case, its features-map entry, the GetFeatureString arm, the featureImplicitYield FSComp.txt string, and regenerate the xlf localisation files. The always-on SynExpr.SequentialOrImplicitYield AST machinery is untouched. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> --------- Co-authored-by: Copilot <copilot@github.com> Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> (cherry picked from commit 9f19d64) * Import: Don't walk non-F# assemblies when labelling trait constraint sources (#20090) * Don't walk non-F# assemblies when labelling trait constraint sources `addConstraintSources` (added in #16304, so that a failed member constraint names the member it came from) is applied to every imported assembly, and recurses through `e.ModuleOrNamespaceType` for every module and namespace entity it finds. For an assembly imported from IL there is nothing to find: the walk only reads `AllValsAndMembers`, and `ImportILTypeDefs` gives every namespace and type entity an empty val list; only an F# trait constraint produces a `TyparConstraint.MayResolveMember` to label in the first place. Meanwhile the recursion forces each namespace entity's `ModuleOrNamespaceType`, which imports that namespace - so referencing an assembly ends up importing every namespace in it, and reading every type definition, whether or not the code touches it. Skip the CCUs that aren't F#. FSharp.Core and F# references are still walked, so the error messages are unchanged. Measured with FSharpChecker.ParseAndCheckProject, keeping the results alive so the imported assembly structures stay on the heap (averages of 3 runs, one per process): a 486-reference F# project retains 1319.2 -> 952.1 MB (-27.8%), and a 168-reference console project 77.7 -> 69.8 MB (-10.2%). Checking FSharp.Compiler.Service itself (124 references, 397 sources) goes 2301.7 -> 2298.0 MB, i.e. within the noise at that size - what the imports cost there is dwarfed by the trees of the project's own code. (cherry picked from commit 623d80d) * chore(Async.RunSynchronouslyImmediate): Stragglers re #19804 (#20245) * chore(Async.RunSynchronouslyImmediate): Stragglers from #19804 - rename and sync clone impls withing VisualFSharp.slnx as per previous PR - update RunImmediateExceptOnUI to delegate and follow naming (cherry picked from commit 1852df8) * Parser: recover on unfinished abstract members (#20070) (cherry picked from commit ce9b62a) * Remove always-on SingleUnderscorePattern language feature flag (#20222) * Remove always-on SingleUnderscorePattern language feature flag Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Remove orphaned parsUnexpectedSymbolDot diagnostic resource Its only two consumers were the parser guards removed when the always-on SingleUnderscorePattern language feature was deleted, leaving the FSComp.txt entry and 13 xlf trans-units unreachable. Regenerated xlf via UpdateXlf. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> --------- Co-authored-by: Copilot <copilot@github.com> Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Co-authored-by: Tomas Grosup <tomas.grosup@microsoft.com> (cherry picked from commit 22e1934) * Add symbol and type highlighting to F# diagnostics (#20097) (cherry picked from commit 208b7b4) * Avoid per-instance lock object in InterruptibleLazy and DelayInitArrayMap (#20088) * Avoid per-instance lock object in InterruptibleLazy and DelayInitArrayMap Both types allocated a dedicated `syncObj = obj()` for their one-time initialisation. These instances are internal and never locked externally, and there are enough of them (one per lazy IL member, per ILTypeDefs / ILMethodDefs, etc.) that the extra bare System.Object adds up to tens of MB on a large project. Lock on `this` instead and drop the field. Measured on a single-file FCS check against a project with ~486 references: bare System.Object instances dropped from ~1,000,000 to ~29,000 (~-22 MB). (cherry picked from commit 29718dd) * Regression matrix: build unmodified repos against local FSharp.Core (#20240) (cherry picked from commit ffa6ef1) * IL: add ILPreNamespace, make ILPreTypeDef creation lazy (#20092) * IL: add ILPreNamespace, make ILPreTypeDef creation lazy (cherry picked from commit d3fab44) * feat(Async): Add exception-unwrapping Await (#19785) (cherry picked from commit 1e94cc5) * IL: fix leaking binary view (#20250) (cherry picked from commit 6ac056a) * IL: use empty tables for members when possible (#20249) (cherry picked from commit e09605c) * Add regression test: #14454, IAsyncDisposable use in task CE in FSI (#20251) (cherry picked from commit 5d6e4dc) * Teach compiler driver to ignore unknown warning codes (#20246) (cherry picked from commit 0640920) * Remove always-on RelaxWhitespace language feature flag (#20226) (cherry picked from commit d26c842) * Add nullness regression test for recursive inline SRTP memoization (#20247) (cherry picked from commit 492d417) * Ship FSharp.Core with a net10.0 target framework (selective port of #20229) Adapted from upstream PR #20229 for the 10.0.4xx servicing packaging model. Ported: FSharpCoreShippedNetTargetFramework=net10.0 knob (Directory.Build.props/.targets, eng/TargetFrameworks.props), FSharp.Core fsproj/nuspec net10.0 asset shipment, FSharp.Core/*.fs(i) #if NETSTANDARD2_1 || NET widening, FSharp.Compiler.fsproj dependent-project TFM wiring, VerifyNetPackage.fsx package validation, ILVerify net10.0 baselines, and FSharp.Core/FCS surface-area test updates. Not ported: nothing was dropped from the upstream patch's actual net10.0 packaging content; the only upstream hunk excluded was an unrelated Central-Package-Management NU1507 NoWarn block that only applies when ManagePackageVersionsCentrally is set (introduced by the separately excluded #20084 CPM migration, not by #20229 itself). No Arcade 11 or .NET 11 SDK/TFM assumptions are introduced; the destination's existing SDK pin, Arcade version, and servicing branding are preserved. (cherry picked from commit 1928203, adapted) * Flow release/10.0.4xx into rel/d18.11-ltsc for VS insertion (adapted from #20412) Upstream PR #20412 ("Make release/dev18.11 flow into rel/insiders") retargeted release/dev18.11's VS insertion branch from rel/main to rel/insiders, and set FSharpReleaseBranchName from main to release/dev18.11. This branch (release/10.0.4xx) already has FSharpReleaseBranchName set to its own servicing branch (release/10.0.4xx), so that part does not apply here and is left untouched. The equivalent, required change for this servicing branch is only the VS insertion target: route VS insertion to rel/d18.11-ltsc instead of rel/insiders, matching the LTSC servicing relationship between this F# branch and its VS branch. No other branch, package, or servicing branding changed. (cherry picked from commit ac4fecf, adapted) * Fix ILVerify baseline filenames for net10.0 FSharp.Core asset (follow-up to #20229 adaptation) The original upstream #20229 commit shipped the ILVerify baseline files named with a literal 'netcoreapp' TFM segment (ilverify_FSharp.Core_{Debug,Release}_netcoreapp.bsl), but ilverify.ps1 resolves the baseline filename dynamically from the FSharpCoreShippedNetTargetFramework MSBuild property, which in this destination resolves to 'net10.0' (not 'net11.0' as in later upstream history, and not the literal string 'netcoreapp'). Renamed the two empty (zero-error) baseline files to ilverify_FSharp.Core_{Debug,Release}_net10.0.bsl so tests/ILVerify/ilverify.ps1 finds them. Verified directly via 'dotnet ilverify --sanity-checks --tokens' against the built Release net10.0 FSharp.Core.dll: 'All Classes and Methods ... Verified.' with zero errors, matching the (empty) baseline. * Use the 43.13 FCS family for the 18.11 backport Signal the upstream breaking tokenizer, tooltip and record AST contracts with a new FCS package family. Keep compiler and Core servicing versions unchanged. Add a red-green version guard and package-linked migration notes. Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com> Copilot-Session: ee897a29-d64c-4e85-954e-cded7192eea6 --------- Co-authored-by: Nat Elkins <nat@nelknet.com> Co-authored-by: dotnet bot <dotnet-bot@dotnetfoundation.org> Co-authored-by: Copilot <copilot@github.com> Co-authored-by: Eugene Auduchinok <eugene.auduchinok@jetbrains.com> Co-authored-by: Copilot <198982749+Copilot@users.noreply.github.com> Co-authored-by: kerams <kerams@users.noreply.github.com> Co-authored-by: Charles Roddie <charles.roddie@cantab.net> Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> Co-authored-by: Adam Boniecki <20281641+abonie@users.noreply.github.com> Co-authored-by: Brian Rourke Boll <brianrourkeboll@users.noreply.github.com> Co-authored-by: Charles Roddie <charles.roddie@summatic.co.uk> Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Co-authored-by: Ruben Bartelink <ruben@bartelink.com> Co-authored-by: Edgar Gonzalez <egonzalez@totallymoney.com> Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com> Co-authored-by: Kirtikumar Anandrao Ramchandani <33368817+KirtiRamchandani@users.noreply.github.com> Co-authored-by: Martin <29605222+Martin521@users.noreply.github.com> Co-authored-by: Tomas Grosup <tomas.grosup@microsoft.com> Copilot-Session: 9d20cd22-45dc-411f-9163-185fd6dd54d9 Copilot-Session: ee897a29-d64c-4e85-954e-cded7192eea6 Copilot-Session: 34374e7f-b8e0-4d82-89f8-d289e6cdd7f6
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
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
The PR Tooling Safety Check runs a separate threat-detection LLM that only sees the workflow description plus the agent's output, not the process steps. When the agent correctly applies
AI-Tooling-Check-Bypassedto a non-fork PR, the detector misreads the bypass label as the agent being manipulated into skipping its scan and raises a false prompt-injection alarm, aborting the run's label and memory outputs.threat-detection.promptnow tells the detector that bypass-labeling non-fork PRs is the designed path defined intooling-check-repo-rules.md, not an injection.