Skip to content

DO NOT MERGE SONARJAVA-7339 Test the SemSitter frontend with PVF - #6548

Draft
rombirli wants to merge 5 commits into
rombirli/semsitter-frontend-switchfrom
rombirli/semsitter-frontend-default-on
Draft

rombirli wants to merge 5 commits into
rombirli/semsitter-frontend-switchfrom
rombirli/semsitter-frontend-default-on

Conversation

@rombirli

@rombirli rombirli commented Nov 20, 2026 •

Copy link
Copy Markdown
Contributor

Stacked on #6547. Do not merge.

Flips the default of the sonar-java-semsitter-frontend flag to on, so any analysis that knows a SemSitter executable (sonar.java.semsitter.executable or SEMSITTER_ANALYZER) parses with SemSitter. sonar.java.semsitter.enabled=false or the flag set to off still brings back ECJ. Without an executable nothing changes.

Used to exercise the SemSitter frontend end to end on CI and on ruling once the server binary is provided.


Summary by Gitar

  • Flag default:
    • Flipped FLAG_DEFAULT to true so SemSitter is used whenever an executable is known
  • Bundled server:
    • Added BundledServer to use the Linux SemSitter server when no executable is configured
  • Parser:
    • Translation now runs SemSitterTranslator::translate and reports syntax errors with line and column
  • Build:
    • Raised requireFilesSize maxsize in sonar-java-plugin/pom.xml to 50000000 to fit the bundled server
  • Enum constants:
    • Kept enum constant constructors and annotation values as ECJ provides them in AnnotationValueImpl

This will update automatically on new commits.

public static final String EXECUTABLE_ENVIRONMENT = "SEMSITTER_ANALYZER";
public static final String CLIENT_ID_ENVIRONMENT = "SONAR_JAVA_LAUNCHDARKLY_CLIENT_ID";
static final boolean FLAG_DEFAULT = false;
static final boolean FLAG_DEFAULT = true;

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

💡 Quality: Failure log and Javadoc still say the default frontend is used

Now that FLAG_DEFAULT is true, isFlagOn returns true when LaunchDarkly can't be reached (line 90), so SemSitter is the frontend that runs. The warning on line 89 still says "the default frontend is used", which is the opposite of what happens. The class Javadoc (line 35) also still says "Anything that goes wrong while asking leaves the default frontend in use". Anyone reading the logs or the docs during a ruling or CI run will get the wrong idea about which parser handled the analysis. The test name without_property_nor_client_id_the_default_frontend_stays (test line 111) has the same problem, since with no client id the provider now turns SemSitter on. Fix: build the log message and Javadoc from FLAG_DEFAULT, or reword them to be neutral, and rename the test.

Make the warning match what actually happens, whatever the default value is:

LOG.warn("The SemSitter frontend flag could not be read, it defaults to {}: {}", FLAG_DEFAULT ? "on" : "off", e.getMessage());
  • Apply fix

Check the box to apply the fix or reply for a change | Was this helpful? React with 👍 / 👎

@datadog-sonarsource

This comment has been minimized.

@rombirli
rombirli added this pull request to stack #6543 November 20, 2026 10:19
@rombirli
rombirli force-pushed the rombirli/semsitter-frontend-default-on branch from 088ea84 to d79893d Compare November 20, 2026 10:29
@rombirli
rombirli force-pushed the rombirli/semsitter-frontend-default-on branch from d79893d to 6bba146 Compare November 20, 2026 10:58
Comment on lines +670 to +672
if (name.endsWith(".*")) {
return declaredIn(symbol, qualifier.replace('$', '.'), importTree.isStatic());
}

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

⚠️ Bug: Static on-demand import flagged unused when only inherited members used

With this PR the SemSitter frontend is on by default, so CompilerWarnings now produces the UNUSED_IMPORT warnings that S1128 reads. For a single static import, uses() calls isMemberOf, which also accepts members the imported type inherits. For a static on-demand import it calls declaredIn, which only accepts a member whose owner is exactly the imported type (line 717). Take import static org.mockito.Mockito.*; in a file that only calls any() or eq(). Those methods are declared in ArgumentMatchers, which Mockito extends, so the import gets a false 'never used' warning. ECJ does not report this. Fix: in the static, non-type branch of declaredIn, use the same inheritance-aware check as isMemberOf.

Reuse the inheritance-aware member check for static on-demand imports (make declaredIn an instance method).:

private boolean declaredIn(Symbol symbol, String qualifier, boolean isStatic) {
  if (symbol.isUnknown()) {
    return false;
  }
  if (isStatic) {
    if (symbol.isTypeSymbol()) {
      String type = dottedType(symbol);
      return type.startsWith(qualifier + ".") && type.indexOf('.', qualifier.length() + 1) < 0;
    }
    return isMemberOf(symbol, qualifier);
  }
  • Apply fix

Check the box to apply the fix or reply for a change | Was this helpful? React with 👍 / 👎

Comment on lines +719 to +723
return symbol.owner() instanceof Symbol.TypeSymbol owner && dottedType(owner).equals(qualifier);
}
Symbol type = symbol;
if ("<init>".equals(symbol.name())) {
type = symbol.owner();

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

💡 Bug: Type-import-on-demand counted as used by types from subpackages

For a non-static on-demand import, declaredIn treats the import as used whenever the type's dotted name starts with qualifier + "." (line 723). It never checks that the type sits directly in that package or type, while the static branch above it does (line 715). So import java.util.*; import java.util.concurrent.ConcurrentHashMap; with only new ConcurrentHashMap<>() in the body marks java.util.* as used: java.util.concurrent.ConcurrentHashMap starts with java.util.. The unused import is then never reported, and S1128 misses an issue ECJ would raise. Fix: also require that no further . follows the qualifier, as the static branch does.

Only count types declared directly in the imported package/type.:

String dotted = dottedType(type);
return type.isTypeSymbol() && !type.isUnknown() && dotted.startsWith(qualifier + ".") && dotted.indexOf('.', qualifier.length() + 1) < 0;
  • Apply fix

Check the box to apply the fix or reply for a change | Was this helpful? React with 👍 / 👎

Comment on lines +133 to +147
@Override
public final boolean equals(Object obj) {
if (this == obj) {
return true;
}
if (!(obj instanceof JSymbol other) || kind != other.kind) {
return false;
}
if (id != null && other.id != null) {
return id.equals(other.id);
}
return name.equals(other.name) && Objects.equals(owner, other.owner);
}

@Override

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

💡 Bug: JSymbol equals/hashCode contract violated for id-less symbols

When one of the two symbols has no id, JSymbol.equals compares them by name and owner (lines 141-144). hashCode still hashes id when it is present and name otherwise (line 150). Production code does create id-less symbols, for example the unresolved JMethodSymbol(sema, null, name) in SemSitterTranslator line 2241 and the sema's unknownSymbol. An id-less symbol can therefore equal a symbol whose id is set while having a different hash code. That breaks the Object.hashCode contract and gives inconsistent results in HashMap/HashSet lookups. Fix: base hashCode only on fields that equals always compares, such as kind and name.

Hash only on fields equals always agrees on.:

hash = Objects.hash(kind, name);
  • Apply fix

Check the box to apply the fix or reply for a change | Was this helpful? React with 👍 / 👎

@rombirli
rombirli force-pushed the rombirli/semsitter-frontend-default-on branch from 6bba146 to b655d0a Compare November 20, 2026 11:25
@rombirli rombirli changed the title DO NOT MERGE SONARJAVA-7339 Use the SemSitter frontend by default DO NOT MERGE SONARJAVA-7339 Test the SemSitter frontend with PVF Nov 20, 2026
@rombirli

Copy link
Copy Markdown
Contributor Author

/pvf

@hashicorp-vault-sonar-prod

Copy link
Copy Markdown
Contributor

📊 Performance Validation complete
🆕 71268 new · 🗑️ 1058705 lost issues
View dashboard

Projects with unexpected PVF results:

  • com.redhat.lightblue-lightblue-core-pom
  • es.usc.citius.hipster-hipster-pom
  • gatherers4j
  • guava
  • guava_java9
  • kafka
  • netflix-conductor
  • org.activiti-activiti-root
  • org.apache.wicket-wicket-parent
  • redisson
  • rxjava

…ature flag

The translator turns the SemSitter documents into the Java tree the checks run on. The flag keeps the default frontend until it turns on.
…xecutable is known

The flag default flips to on, so an analysis with a SemSitter executable parses with SemSitter
unless the flag or sonar.java.semsitter.enabled turns it off.
…uses that frontend

The performance validation runners have no SemSitter executable. This build ships
the protocol 2 Linux server and turns the frontend on when no path is configured.
…ing PVF

Other Linux analyses, including ruling, stay on the default frontend.
@rombirli
rombirli force-pushed the rombirli/semsitter-frontend-default-on branch from 439646f to 4db55e1 Compare November 20, 2026 15:48
@gitar-bot

gitar-bot Bot commented Nov 20, 2026 •

Copy link
Copy Markdown
CI failed: The ruling-update job failed because its prerequisite ruling QA job was cancelled, so no artifacts were downloaded and the sync step errored. The analyzer could not tie the upstream cancellation to this PR, so the failure is attributed to neither the change nor a confirmed cause.

Overview

The ruling-update job failed because its prerequisite ruling QA job was cancelled, so no artifacts were downloaded and the sync step errored. The analyzer could not tie the upstream cancellation to this PR, so the failure is attributed to neither the change nor a confirmed cause.

Failures

Ruling-update job failed after upstream ruling QA was cancelled, leaving no artifacts to sync

  • Type: configuration
  • Affected jobs: 136394005890
  • Related to change: unclear
  • Confidence: high
  • Root cause: The log shows needs.ruling-qa.result was cancelled; the update script consequently set ruling-failed=true. Artifact download found 0 artifacts, and the sync step failed because the expected artifact directory did not exist. This is a downstream consequence of the cancellation, not evidence that the ruling checks themselves failed. The PR changes the SemSitter frontend for PVF, but the logs do not explain why ruling QA was cancelled or identify a concrete change that caused it, so the relationship is unclear.
  • Suggested fix: Handle cancelled separately from failure in the ruling-update workflow: skip artifact synchronization and notification when ruling QA was cancelled, or ensure the prerequisite completes and uploads artifacts before running the update job. Investigate the upstream cancellation reason and rerun ruling QA if it was unintended.

Summary

  • Change-related failures: 0
  • Infrastructure/flaky failures: 0
  • Recommended action: Handle cancelled separately from failure in the ruling-update workflow: skip artifact synchronization and notification when ruling QA was cancelled, or ensure the prerequisite completes and uploads artifacts before running the update job. Investigate the upstream cancellation reason and rerun ruling QA if it was unintended.
Code Review ⚠️ Changes requested 🔴 High risk 0 closed / 4 findings

Defaulting analyses with a SemSitter executable to the new translator could change parsing and findings across affected Java projects.

Flips the SemSitter frontend flag default to on so analyses with a known SemSitter executable parse with SemSitter, with the exception of sonar.java.semsitter.enabled=false. Changes requested: the failure log, Javadoc, and test name in SemSitterParserProvider still describe ECJ as the default frontend, which is now wrong. CompilerWarnings also has open issues: static on-demand imports are wrongly flagged as unused when only inherited members are used, and type on-demand imports are wrongly counted as used by types in subpackages. JSymbol.hashCode is inconsistent with equals for id-less symbols and must hash only on kind and name.

⚠️ Bug: Static on-demand import flagged unused when only inherited members used

📄 java-frontend-semsitter/src/main/java/org/sonar/java/semsitter/CompilerWarnings.java:670-672 📄 java-frontend-semsitter/src/main/java/org/sonar/java/semsitter/CompilerWarnings.java:676-685 📄 java-frontend-semsitter/src/main/java/org/sonar/java/semsitter/CompilerWarnings.java:710-718 📄 java-frontend-semsitter/src/main/java/org/sonar/java/semsitter/CompilerWarnings.java:566-568

With this PR the SemSitter frontend is on by default, so CompilerWarnings now produces the UNUSED_IMPORT warnings that S1128 reads. For a single static import, uses() calls isMemberOf, which also accepts members the imported type inherits. For a static on-demand import it calls declaredIn, which only accepts a member whose owner is exactly the imported type (line 717). Take import static org.mockito.Mockito.*; in a file that only calls any() or eq(). Those methods are declared in ArgumentMatchers, which Mockito extends, so the import gets a false 'never used' warning. ECJ does not report this. Fix: in the static, non-type branch of declaredIn, use the same inheritance-aware check as isMemberOf.

Reuse the inheritance-aware member check for static on-demand imports (make declaredIn an instance method).
private boolean declaredIn(Symbol symbol, String qualifier, boolean isStatic) {
  if (symbol.isUnknown()) {
    return false;
  }
  if (isStatic) {
    if (symbol.isTypeSymbol()) {
      String type = dottedType(symbol);
      return type.startsWith(qualifier + ".") && type.indexOf('.', qualifier.length() + 1) < 0;
    }
    return isMemberOf(symbol, qualifier);
  }
💡 Quality: Failure log and Javadoc still say the default frontend is used

📄 java-frontend-semsitter/src/main/java/org/sonar/java/semsitter/SemSitterParserProvider.java:33-35 📄 java-frontend-semsitter/src/main/java/org/sonar/java/semsitter/SemSitterParserProvider.java:45 📄 java-frontend-semsitter/src/main/java/org/sonar/java/semsitter/SemSitterParserProvider.java:89-90 📄 java-frontend-semsitter/src/test/java/org/sonar/java/semsitter/SemSitterParserProviderTest.java:111-116 📄 java-frontend-semsitter/src/test/java/org/sonar/java/semsitter/SemSitterParserProviderTest.java:114-120 📄 java-frontend-semsitter/src/main/java/org/sonar/java/semsitter/SemSitterParserProvider.java:47 📄 java-frontend-semsitter/src/main/java/org/sonar/java/semsitter/SemSitterParserProvider.java:87-89

Now that FLAG_DEFAULT is true, isFlagOn returns true when LaunchDarkly can't be reached (line 90), so SemSitter is the frontend that runs. The warning on line 89 still says "the default frontend is used", which is the opposite of what happens. The class Javadoc (line 35) also still says "Anything that goes wrong while asking leaves the default frontend in use". Anyone reading the logs or the docs during a ruling or CI run will get the wrong idea about which parser handled the analysis. The test name without_property_nor_client_id_the_default_frontend_stays (test line 111) has the same problem, since with no client id the provider now turns SemSitter on. Fix: build the log message and Javadoc from FLAG_DEFAULT, or reword them to be neutral, and rename the test.

Make the warning match what actually happens, whatever the default value is
LOG.warn("The SemSitter frontend flag could not be read, it defaults to {}: {}", FLAG_DEFAULT ? "on" : "off", e.getMessage());
💡 Bug: Type-import-on-demand counted as used by types from subpackages

📄 java-frontend-semsitter/src/main/java/org/sonar/java/semsitter/CompilerWarnings.java:719-723

For a non-static on-demand import, declaredIn treats the import as used whenever the type's dotted name starts with qualifier + "." (line 723). It never checks that the type sits directly in that package or type, while the static branch above it does (line 715). So import java.util.*; import java.util.concurrent.ConcurrentHashMap; with only new ConcurrentHashMap<>() in the body marks java.util.* as used: java.util.concurrent.ConcurrentHashMap starts with java.util.. The unused import is then never reported, and S1128 misses an issue ECJ would raise. Fix: also require that no further . follows the qualifier, as the static branch does.

Only count types declared directly in the imported package/type.
String dotted = dottedType(type);
return type.isTypeSymbol() && !type.isUnknown() && dotted.startsWith(qualifier + ".") && dotted.indexOf('.', qualifier.length() + 1) < 0;
💡 Bug: JSymbol equals/hashCode contract violated for id-less symbols

📄 java-frontend-semsitter/src/main/java/org/sonar/java/semsitter/JSymbol.java:133-147 📄 java-frontend-semsitter/src/main/java/org/sonar/java/semsitter/SemSitterTranslator.java:2241

When one of the two symbols has no id, JSymbol.equals compares them by name and owner (lines 141-144). hashCode still hashes id when it is present and name otherwise (line 150). Production code does create id-less symbols, for example the unresolved JMethodSymbol(sema, null, name) in SemSitterTranslator line 2241 and the sema's unknownSymbol. An id-less symbol can therefore equal a symbol whose id is set while having a different hash code. That breaks the Object.hashCode contract and gives inconsistent results in HashMap/HashSet lookups. Fix: base hashCode only on fields that equals always compares, such as kind and name.

Hash only on fields equals always agrees on.
hash = Objects.hash(kind, name);
🤖 Prompt for agents
Code Review: Flips the SemSitter frontend flag default to on so analyses with a known SemSitter executable parse with SemSitter, with the exception of `sonar.java.semsitter.enabled=false`. Changes requested: the failure log, Javadoc, and test name in `SemSitterParserProvider` still describe ECJ as the default frontend, which is now wrong. `CompilerWarnings` also has open issues: static on-demand imports are wrongly flagged as unused when only inherited members are used, and type on-demand imports are wrongly counted as used by types in subpackages. `JSymbol.hashCode` is inconsistent with `equals` for id-less symbols and must hash only on `kind` and `name`.

1. 💡 Quality: Failure log and Javadoc still say the default frontend is used
   Files: java-frontend-semsitter/src/main/java/org/sonar/java/semsitter/SemSitterParserProvider.java:33-35, java-frontend-semsitter/src/main/java/org/sonar/java/semsitter/SemSitterParserProvider.java:45, java-frontend-semsitter/src/main/java/org/sonar/java/semsitter/SemSitterParserProvider.java:89-90, java-frontend-semsitter/src/test/java/org/sonar/java/semsitter/SemSitterParserProviderTest.java:111-116, java-frontend-semsitter/src/test/java/org/sonar/java/semsitter/SemSitterParserProviderTest.java:114-120, java-frontend-semsitter/src/main/java/org/sonar/java/semsitter/SemSitterParserProvider.java:47, java-frontend-semsitter/src/main/java/org/sonar/java/semsitter/SemSitterParserProvider.java:87-89

   Now that `FLAG_DEFAULT` is `true`, `isFlagOn` returns `true` when LaunchDarkly can't be reached (line 90), so SemSitter is the frontend that runs. The warning on line 89 still says "the default frontend is used", which is the opposite of what happens. The class Javadoc (line 35) also still says "Anything that goes wrong while asking leaves the default frontend in use". Anyone reading the logs or the docs during a ruling or CI run will get the wrong idea about which parser handled the analysis. The test name `without_property_nor_client_id_the_default_frontend_stays` (test line 111) has the same problem, since with no client id the provider now turns SemSitter on. Fix: build the log message and Javadoc from `FLAG_DEFAULT`, or reword them to be neutral, and rename the test.

   Fix (Make the warning match what actually happens, whatever the default value is):
   LOG.warn("The SemSitter frontend flag could not be read, it defaults to {}: {}", FLAG_DEFAULT ? "on" : "off", e.getMessage());

2. ⚠️ Bug: Static on-demand import flagged unused when only inherited members used
   Files: java-frontend-semsitter/src/main/java/org/sonar/java/semsitter/CompilerWarnings.java:670-672, java-frontend-semsitter/src/main/java/org/sonar/java/semsitter/CompilerWarnings.java:676-685, java-frontend-semsitter/src/main/java/org/sonar/java/semsitter/CompilerWarnings.java:710-718, java-frontend-semsitter/src/main/java/org/sonar/java/semsitter/CompilerWarnings.java:566-568

   With this PR the SemSitter frontend is on by default, so `CompilerWarnings` now produces the UNUSED_IMPORT warnings that S1128 reads. For a single static import, `uses()` calls `isMemberOf`, which also accepts members the imported type inherits. For a static on-demand import it calls `declaredIn`, which only accepts a member whose owner is exactly the imported type (line 717). Take `import static org.mockito.Mockito.*;` in a file that only calls `any()` or `eq()`. Those methods are declared in `ArgumentMatchers`, which `Mockito` extends, so the import gets a false 'never used' warning. ECJ does not report this. Fix: in the static, non-type branch of `declaredIn`, use the same inheritance-aware check as `isMemberOf`.

   Fix (Reuse the inheritance-aware member check for static on-demand imports (make declaredIn an instance method).):
   private boolean declaredIn(Symbol symbol, String qualifier, boolean isStatic) {
     if (symbol.isUnknown()) {
       return false;
     }
     if (isStatic) {
       if (symbol.isTypeSymbol()) {
         String type = dottedType(symbol);
         return type.startsWith(qualifier + ".") && type.indexOf('.', qualifier.length() + 1) < 0;
       }
       return isMemberOf(symbol, qualifier);
     }

3. 💡 Bug: Type-import-on-demand counted as used by types from subpackages
   Files: java-frontend-semsitter/src/main/java/org/sonar/java/semsitter/CompilerWarnings.java:719-723

   For a non-static on-demand import, `declaredIn` treats the import as used whenever the type's dotted name starts with `qualifier + "."` (line 723). It never checks that the type sits directly in that package or type, while the static branch above it does (line 715). So `import java.util.*; import java.util.concurrent.ConcurrentHashMap;` with only `new ConcurrentHashMap<>()` in the body marks `java.util.*` as used: `java.util.concurrent.ConcurrentHashMap` starts with `java.util.`. The unused import is then never reported, and S1128 misses an issue ECJ would raise. Fix: also require that no further `.` follows the qualifier, as the static branch does.

   Fix (Only count types declared directly in the imported package/type.):
   String dotted = dottedType(type);
   return type.isTypeSymbol() && !type.isUnknown() && dotted.startsWith(qualifier + ".") && dotted.indexOf('.', qualifier.length() + 1) < 0;

4. 💡 Bug: JSymbol equals/hashCode contract violated for id-less symbols
   Files: java-frontend-semsitter/src/main/java/org/sonar/java/semsitter/JSymbol.java:133-147, java-frontend-semsitter/src/main/java/org/sonar/java/semsitter/SemSitterTranslator.java:2241

   When one of the two symbols has no id, `JSymbol.equals` compares them by name and owner (lines 141-144). `hashCode` still hashes `id` when it is present and `name` otherwise (line 150). Production code does create id-less symbols, for example the unresolved `JMethodSymbol(sema, null, name)` in `SemSitterTranslator` line 2241 and the sema's `unknownSymbol`. An id-less symbol can therefore equal a symbol whose id is set while having a different hash code. That breaks the `Object.hashCode` contract and gives inconsistent results in `HashMap`/`HashSet` lookups. Fix: base `hashCode` only on fields that `equals` always compares, such as `kind` and `name`.

   Fix (Hash only on fields equals always agrees on.):
   hash = Objects.hash(kind, name);

Verification Rules 2/2 met Repos 5 inspected · PRs 2 inspected

Functional validation No results

Rules 2 passed 2 not applicable

✅ Internal artifact registries
✅ Repository license compliance

Broader impact Repos 5 inspected PRs 2 inspected
Cross-repo 5 repositories inspected 0 open impacts

SonarSource/sonar-iac Inspected
SonarSource/sonarlint-vscode Inspected
SonarSource/sonar-lits Inspected
SonarSource/sonarlint-intellij Inspected
SonarSource/sonarlint-eclipse Inspected

Cross-PR 2 PRs inspected 0 open impacts

PR 6547: SONARJAVA-7339 Make the Java frontend switchable between ECJ and SemSitter Inspected
PR 6542: SONARJAVA-7339 Add the SemSitter server client Inspected

Auto-approval Not enabled

Set up

For more verification details, comment gitar display:verbose.

Tip

Comment Gitar fix CI or enable auto-apply: gitar auto-apply:on

Options

Auto-apply is off → Gitar will not commit updates to this branch.
Display: compact → Showing compact verification results.
Unblock → Override a blocking verdict and allow merging.

Comment with these commands to change the behavior for this request:

Auto-apply Compact Unblock
gitar auto-apply:on         
gitar display:verbose         
gitar unblock         

Was this helpful? React with 👍 / 👎 | Gitar

@rombirli

Copy link
Copy Markdown
Contributor Author

/pvf

@sonarqube-next

Copy link
Copy Markdown
Contributor

Quality Gate failed Quality Gate failed

Failed conditions
308 New issues
25.3% Coverage on New Code (required ≥ 90%)

See analysis details on SonarQube

Catch issues before they fail your Quality Gate with our IDE extension SonarQube for IDE SonarQube for IDE

@hashicorp-vault-sonar-prod

Copy link
Copy Markdown
Contributor

📊 Performance Validation complete
🆕 12055 new · 🗑️ 403740 lost issues
View dashboard

Projects with unexpected PVF results:

  • org.apache.jackrabbit-jackrabbit
  • tomcat8

This branch has not been deployed

No deployments
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant