When using `ProjectMatrix` with `CrossVersion.full` and Scala 2 versions like `2.13.18`, the project ID incorrectly became `$1$2_13_18` instead of `foo2_13_18`.
**Root cause:** The Scala 3 compiler creates synthetic intermediate vals (e.g., `$1`) during macro expansion. The `enclosingTerm` function in the macros was stopping at these synthetic symbols instead of continuing up the symbol tree to find the actual val name.
**Fix:** Added `Flags.Synthetic` check to skip compiler-generated symbols in:
- `main-settings/src/main/scala/sbt/std/KeyMacro.scala`
- `lm-core/src/main/scala/sbt/librarymanagement/ConfigurationExtra.scala`
---------
Co-authored-by: byteforge38 <joseph.mc0803@gmail.com>
When Java compiler generates warnings about missing annotations from
JAR files, the path format is jar:file:///C:/... which causes
InvalidPathException on Windows due to the : character.
The fix filters out jar: paths in toDocument(), similar to how fake
positions like <macro> are already filtered out. This prevents the
exception and allows compilation to continue.
Diagnostics for files inside JARs are not shown in the IDE, which is
correct behavior since they cannot be edited.
Fixes#7665
Generated-by: Cascade (AI pair programmer)
Fixes#7826
Changes:
- Upgrade sbt version from 0.13.18 to 1.10.7 in launcher-package/project/build.properties
- Remove dead dispatch-http 0.8.10 dependency from plugins.sbt
- Update sbt-native-packager from 1.7.6 to 1.10.4 (com.github.sbt)
- Update sbt-pgp from 2.1.2 to 2.3.1
- Replace all dispatch.classic._ HTTP calls with Java URL.openStream()
The dispatch-http library is dead and doesn't support sbt 1.x. This change
uses standard Java networking APIs for HTTP downloads, eliminating the
dependency on outdated libraries.
All HTTP download operations now use:
- java.net.URL.openConnection()
- InputStream with buffered reading
- Proper resource cleanup with try-finally blocks
- Removes the sbt-export-repo plugin dependency
- Defines exportRepo and exportRepoDirectory settings locally
- Removes ExportRepoPlugin from dist project
- Fixes exportRepo task to not reference itself
* fix: Update PackageSignerPlugin.scala for sbt 1.x compatibility
- Use withExtension instead of copy(extension = ...) for Artifact
- Update Classpaths.publishConfig to use sbt 1.x signature with all required parameters
- Use new slash syntax for scoped keys (pgpSigner / skip instead of skip in pgpSigner)
* fix: Use computed status instead of status key in PackageSignerPlugin
* fix: Use withRevision instead of copy for ModuleID in build.sbt
- Add scala.sys.process._ import for shell command execution (! method)
- Fix resolvers setting to use new slash syntax and proper Option handling
- Fix publish/publishLocal to use {} instead of () to avoid deprecation warning
- Fix Classpaths.publishTask to use variant without delivery key (deprecated since 1.1.1)
- Add scala.sys.process._ import for shell command execution
- Fix publish/publishLocal to use {} instead of ()
- Remove problematic resolvers setting that caused task/setting dependency issues
- Use IvyActions.publish directly instead of deprecated Classpaths.publishTask
- Fix ModuleID.copy to use withRevision
- Fix Artifact.copy to use withExtension
- Use computed status value instead of status key
Build now compiles successfully with sbt 1.10.7.
Post-process the delivered Ivy XML file to apply dependency overrides.
This is necessary because Ivy's deliver() method doesn't automatically
apply DependencyDescriptorMediators when writing the XML.
The fix:
- Adds applyOverridesToDeliveredIvy() method in IvyActions
- Reads the delivered Ivy XML and transforms dependency elements
- Updates rev attributes for dependencies that have overrides
- Ensures consistency between Maven POM and Ivy XML publishing
Added unit tests to verify:
- Overrides are correctly applied to matching dependencies
- All other attributes are preserved
- Non-matching dependencies remain unchanged
Address review feedback from eed3si9n to make it clearer that GC time
is cumulative CPU time across parallel collectors, which is why it can
exceed wall clock time.
Fixes#8002
**Problem**
consoleProject doesn't work. REPL doesn't even start.
**Solution**
I made some progress into consoleProject.
At least Scala 3.7 repl session will now start.
The problem is that compiler bridge has not implemented binding,
so we can't forward the sbt build information into the repl.
- Ensure output files are synced even when reading from symlink
- Call findActionResult to get ActionResult for side-effect files
- Maintains performance benefit while ensuring correctness
- Addresses @eed3si9n's concern about tasks generating files on the side
- Replace try-catch with Exception.nonFatalCatch.opt for cleaner code
- Follows Scala best practices for non-fatal exception handling
- More functional and idiomatic approach
- Avoids catching fatal exceptions like VirtualMachineError
- Use config.fileConverter.toPath() instead of string manipulation
- Avoid hardcoded '/' prefix removal
- More robust and maintainable approach
- Fix IO.read() to include UTF-8 charset parameter
**Problems**
When running forked tests, sbt uses `Runtime.getRuntime().availableProcessors()` to determine the thread pool size, ignoring `concurrentRestrictions`. This is inconsistent with non-forked parallel tests.
**Expectations**
Users should be able to control the number of parallel test threads in forked mode, similar to how `concurrentRestrictions` works for non-forked tests.
**Notes**
Added a new setting `testForkedParallelism` that allows explicit control:
```scala
testForkedParallelism := Some(2) // Use 2 threads
testForkedParallelism := None // Use availableProcessors() (default)
```
- Check for symlinked value files before reading AC JSON
- When symlink exists and is valid, read directly from it
- Fallback to original AC file reading if symlink read fails
- Improves performance by avoiding unnecessary AC file reads
- All existing tests pass
xuwei-k
byteforge
MkDev11
eugene yokota
tellorian
gayanMatch
calm329