Stage the replication candidate before the fix phase can time out

Seven of the eight timed-out runs in the cached corpus had already
reproduced their issue, authored a failing test and cleared the negative
control. All seven published nothing, because the candidate manifest was
written after the fix phase and a task timeout kills the process rather
than raising into the try/catch around it.

Extract the manifest into a single $writeCandidateManifest scriptblock
and call it twice: before the fix phase to stage the reproduction, and
after it to announce READY with any fix attached. One writer, so the two
manifests cannot drift apart.

The pipeline guard that cross-checks the gate allowlist against the
fields the orchestrator writes matched top-level keys by a hard-coded
eight-space indent, so the move into a scriptblock emptied its derived
list. Derive the indent from the manifest instead.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Copilot-Session: 735ac9a2-7bec-4baa-ad19-c298e5bc795a

Author: Copilot CI

.github/scripts/Replicate-Issue.Tests.ps1

@@ -13439,3 +13439,59 @@ Describe 'An attempt classifier logs the text it classified' {
             'veto that cannot be attributed to an attempt cannot be audited')
     }
 }
+
+Describe 'A verified reproduction survives a fix-phase timeout' {
+    # Run 15089945 reproduced its issue, passed verification, cleared the
+    # negative control and selected a winning fix - then hit
+    # "##[error]The task has timed out." before the candidate manifest existed,
+    # so the publisher reported "No replication candidate manifest was
+    # produced; nothing to validate." and opened nothing. 7 of the 8 timeouts
+    # in the cached corpus had already passed verification. A task timeout
+    # kills the process, so the existing try/catch around the fix phase cannot
+    # help: the manifest has to be on disk before the panel starts.
+    BeforeAll {
+        $script:ReplSource = Get-Content -Raw -LiteralPath (
+            Join-Path $PSScriptRoot 'Replicate-Issue.ps1')
+        $script:StageAt = $script:ReplSource.IndexOf('& $writeCandidateManifest $false')
+        $script:FixAt = $script:ReplSource.IndexOf('$fixOutcome = Invoke-ReplicationFixPhase')
+        $script:FinalAt = $script:ReplSource.IndexOf('& $writeCandidateManifest $true')
+        $script:DefAt = $script:ReplSource.IndexOf('$writeCandidateManifest = {')
+    }
+
+    It 'writes the manifest before the fix phase starts' {
+        $script:StageAt | Should -BeGreaterThan 0
+        $script:FixAt | Should -BeGreaterThan 0
+        $script:StageAt | Should -BeLessThan $script:FixAt
+    }
+
+    It 'defines the writer above the call that stages it' {
+        # All four sites are top level, so source order is execution order.
+        $script:DefAt | Should -BeGreaterThan 0
+        $script:DefAt | Should -BeLessThan $script:StageAt
+    }
+
+    It 'writes it again after the fix phase, so a fix is not lost' {
+        $script:FinalAt | Should -BeGreaterThan $script:FixAt
+    }
+
+    It 'announces READY only once, after the fix phase' {
+        # The staged write must not claim READY: that marker is what the
+        # corpus census and the publisher key on.
+        ([regex]::Matches($script:ReplSource,
+            'ISSUE REPLICATION CANDIDATE READY')).Count | Should -Be 1
+        $readyAt = $script:ReplSource.IndexOf('ISSUE REPLICATION CANDIDATE READY')
+        $script:ReplSource.Substring($readyAt - 400, 400) | Should -Match '\$Announce'
+    }
+
+    It 'still records a staged manifest so the timeout case stays measurable' {
+        $script:ReplSource | Should -Match 'ISSUE REPLICATION CANDIDATE STAGED'
+    }
+
+    It 'stages the identical manifest the fix-less run would publish' {
+        # Two hand-written manifests would drift, and the drift would only ever
+        # show up as a malformed candidate on the timeout path.
+        ([regex]::Matches($script:ReplSource,
+            '\$writeCandidateManifest = \{')).Count | Should -Be 1
+    }
+}
+

.github/scripts/Replicate-Issue.ps1

@@ -8137,10 +8137,89 @@ Explain in lighterTypesRejected why the previous tier could not observe it. Choo
 
     New-TestPatch -Files $generatedFiles
 
+    # A task timeout kills this process outright, so the guard above -
+    # which only survives the fix phase *failing* - cannot help: run 15089945
+    # reproduced its issue, verified it, cleared the negative control and
+    # picked a winning fix, then timed out before this manifest existed. The
+    # publisher then said "No replication candidate manifest was produced;
+    # nothing to validate." 7 of the 8 timeouts in the cached corpus had
+    # already passed verification, which makes this the most expensive way
+    # the pipeline loses work.
+    #
+    # Writing the manifest once before the fix phase and again after leaves a
+    # valid reproduction-only candidate on disk the whole time the fix panel
+    # runs. The artifact upload is succeededOrFailed, so it survives, and the
+    # manifest already models a fix-less candidate (fixFiles = @()).
+    $writeCandidateManifest = {
+        param([bool]$Announce)
+        # Replacing newlines after truncation left a trailing space, and the gate
+        # rejects an untrimmed manifest step, so build 15030804 reproduced its issue
+        # and was discarded for whitespace. Collapse and trim after the replacement.
+        $reproductionSteps = @($testProposal.reproductionSteps | ForEach-Object {
+            ([regex]::Replace(
+                ((ConvertTo-ReplicationSafeLog $_ 300) -replace '\r|\n', ' '),
+                '\s+',
+                ' ')).Trim()
+        } | Where-Object { -not [string]::IsNullOrWhiteSpace($_) } | Select-Object -First 10)
+        [ordered]@{
+            schemaVersion = 1
+            issueNumber = $IssueNumber
+            platform = $Platform
+            baseSha = $BaseSha.ToLowerInvariant()
+            status = 'reproduced'
+            blocked = $null
+            selectedDevice = [ordered]@{
+                id = $selectedDeviceId
+                name = $DeviceName
+                osVersion = $DeviceOSVersion
+            }
+            attempts = [ordered]@{
+                sandbox = $sandboxAttempts
+                automatedTest = $testAttempts
+            }
+            reproductionSteps = $reproductionSteps
+            expectedBehavior = ConvertTo-ReplicationSafeLog ([string]$testProposal.expectedBehavior) 500
+            observedBehavior = ConvertTo-ReplicationSafeLog ([string]$testProposal.observedBehavior) 500
+            testType = [string]$testProposal.testType
+            testFilter = [string]$testProposal.testFilter
+            testClassName = [string]$verifierMetadata.ClassName
+            testMethodName = [string]$verifierMetadata.MethodName
+            expectedFailureSignature = [string]$testProposal.expectedFailureSignature
+            files = $generatedFiles
+            sandboxFiles = [ordered]@{
+                xaml = 'sandbox/MainPage.xaml'
+                codeBehind = 'sandbox/MainPage.xaml.cs'
+                appiumPlan = 'sandbox/appium-plan.json'
+            }
+            reproductionResult = 'reproduction-result.json'
+            evidenceManifest = 'evidence/evidence.json'
+            verificationResult = 'verification/verification-result.json'
+            negativeControl = $negativeControl
+            patch = 'test.patch'
+            fixFiles = if ($fixOutcome) { @($fixOutcome.Files) } else { @() }
+            fixPatch = if ($fixOutcome) { 'fix.patch' } else { $null }
+            fixRootCause = if ($fixOutcome) { $fixOutcome.RootCause } else { $null }
+            fixApproach = if ($fixOutcome) { $fixOutcome.Approach } else { $null }
+            fixRejectedApproaches = if ($fixOutcome) { @($fixOutcome.RejectedApproaches) } else { @() }
+        } | ConvertTo-Json -Depth 10 | Set-Content -LiteralPath $candidatePath -Encoding utf8NoBOM
+
+        if ($Announce) {
+            Write-Host "ISSUE REPLICATION CANDIDATE READY: $candidatePath"
+        } else {
+            Write-Host "ISSUE REPLICATION CANDIDATE STAGED: $candidatePath"
+        }
+    }
+
+
+    # Stage the reproduction before the fix phase can time out. This writes the
+    # same manifest the run would publish with no fix, so a process killed
+    # during the panel leaves exactly what a fix-less run leaves.
+    $fixOutcome = $null
+    & $writeCandidateManifest $false
+
     # The reproduction is certified at this point and its patch is already
     # written, so nothing the fix phase does can reach it. Any failure inside
     # returns $null and publishes the reproduction alone.
-    $fixOutcome = $null
     if ($negativeControl) {
         try {
             $fixOutcome = Invoke-ReplicationFixPhase `
@@ -8162,58 +8241,7 @@ Explain in lighterTypesRejected why the previous tier could not observe it. Choo
         Remove-Item -LiteralPath $fixPatchPath -Force -ErrorAction SilentlyContinue
     }
 
-    # Replacing newlines after truncation left a trailing space, and the gate
-    # rejects an untrimmed manifest step, so build 15030804 reproduced its issue
-    # and was discarded for whitespace. Collapse and trim after the replacement.
-    $reproductionSteps = @($testProposal.reproductionSteps | ForEach-Object {
-        ([regex]::Replace(
-            ((ConvertTo-ReplicationSafeLog $_ 300) -replace '\r|\n', ' '),
-            '\s+',
-            ' ')).Trim()
-    } | Where-Object { -not [string]::IsNullOrWhiteSpace($_) } | Select-Object -First 10)
-    [ordered]@{
-        schemaVersion = 1
-        issueNumber = $IssueNumber
-        platform = $Platform
-        baseSha = $BaseSha.ToLowerInvariant()
-        status = 'reproduced'
-        blocked = $null
-        selectedDevice = [ordered]@{
-            id = $selectedDeviceId
-            name = $DeviceName
-            osVersion = $DeviceOSVersion
-        }
-        attempts = [ordered]@{
-            sandbox = $sandboxAttempts
-            automatedTest = $testAttempts
-        }
-        reproductionSteps = $reproductionSteps
-        expectedBehavior = ConvertTo-ReplicationSafeLog ([string]$testProposal.expectedBehavior) 500
-        observedBehavior = ConvertTo-ReplicationSafeLog ([string]$testProposal.observedBehavior) 500
-        testType = [string]$testProposal.testType
-        testFilter = [string]$testProposal.testFilter
-        testClassName = [string]$verifierMetadata.ClassName
-        testMethodName = [string]$verifierMetadata.MethodName
-        expectedFailureSignature = [string]$testProposal.expectedFailureSignature
-        files = $generatedFiles
-        sandboxFiles = [ordered]@{
-            xaml = 'sandbox/MainPage.xaml'
-            codeBehind = 'sandbox/MainPage.xaml.cs'
-            appiumPlan = 'sandbox/appium-plan.json'
-        }
-        reproductionResult = 'reproduction-result.json'
-        evidenceManifest = 'evidence/evidence.json'
-        verificationResult = 'verification/verification-result.json'
-        negativeControl = $negativeControl
-        patch = 'test.patch'
-        fixFiles = if ($fixOutcome) { @($fixOutcome.Files) } else { @() }
-        fixPatch = if ($fixOutcome) { 'fix.patch' } else { $null }
-        fixRootCause = if ($fixOutcome) { $fixOutcome.RootCause } else { $null }
-        fixApproach = if ($fixOutcome) { $fixOutcome.Approach } else { $null }
-        fixRejectedApproaches = if ($fixOutcome) { @($fixOutcome.RejectedApproaches) } else { @() }
-    } | ConvertTo-Json -Depth 10 | Set-Content -LiteralPath $candidatePath -Encoding utf8NoBOM
-
-    Write-Host "ISSUE REPLICATION CANDIDATE READY: $candidatePath"
+    & $writeCandidateManifest $true
 }
 catch {
     $rawReason = [string]$_.Exception.Message

.github/scripts/Replication-Pipeline.Tests.ps1

@@ -590,8 +590,19 @@ Describe 'The trusted publisher stages every module its gate loads' {
 
         # Only the manifest's own top-level keys are validated by the gate's
         # allowlist; nested hashtable keys are checked by their own rules.
+        # Derive the top-level indent rather than hard-coding it: the manifest
+        # moved into a scriptblock so it could be written before the fix phase,
+        # which shifted every key four columns and silently emptied this list.
+        $indent = [regex]::Match(
+            $manifestBlock, "(?m)^([ ]*)schemaVersion = 1").Groups[1].Value
+        if (-not $indent) {
+            $lineStart = $orchestrator.LastIndexOf("`n", $manifestStart)
+            $indent = $orchestrator.Substring(
+                $lineStart + 1, $manifestStart - $lineStart - 1)
+        }
+        $indent | Should -Match '^ +$'
         $written = @([regex]::Matches(
-                $manifestBlock, "(?m)^        ([A-Za-z][A-Za-z0-9]*) = ") |
+                $manifestBlock, "(?m)^$indent([A-Za-z][A-Za-z0-9]*) = ") |
             ForEach-Object { $_.Groups[1].Value })
         $written | Should -Contain 'negativeControl'
         $written.Count | Should -BeGreaterThan 10