**Problem**
Some custom LSP calls do not check initialize-handshake,
which over TCP includes token-based authentication.
**Solution**
This adds checkAuthenticated check around sbt/exec etc.
The thin client's blockUntilStart loop recursed while the portfile was
missing and the forked process looked valid, with no deadline: a forked
server that died before writing project/target/active.json, or stayed
alive but wedged, hung the client forever with no output. On Windows the
check ignored process death entirely (Properties.isWin short-circuited
the liveness test), making the hang unconditional there.
The wait is now bounded (default 5 minutes, tunable with
-Dsbt.client.boot.timeout.seconds). On expiry the client fails with a
"did not start within N seconds" message and then prints the server's
captured stderr, instead of hanging silently (#9484). The Windows
liveness workaround is preserved; the deadline is what bounds it.
Generated-by: kimi-code/k3 (Oh My Pi)
**Problem**
There's a race condition between per-byte readSystemIn notification
and one-byte-read thread lifecycle.
Note: One-byte-read thread was introduced as a solution to the problem
that switching the terminal between raw and canonical mode cannot happen
if it's blocked by read.
**Solution**
This eliminates the thread lifecycle issue by keeping the thread alive
throughout the lifecycle of sbtn itself.
read is still called on demand by the server readSystemIn notification.
Running reboot in the sbt shell dropped to the OS shell instead of rebooting.
The break was in teardown: Server.shutdown opened with
log.info, and during a client-initiated reboot the terminal in scope is that
client's already-closed virtual terminal, so the log write throws
ClosedChannelException through the terminal proxy. That aborted teardown
before the portfile was deleted and the server socket closed, and the
exception was swallowed by the shutdown hook (whose own error print goes to
the same dead terminal).
Server.shutdown now completes its state cleanup (portfile, tokenfile, running
flag, server socket) before logging, and CommandExchange.shutdown wraps each
channel shutdown and the server shutdown individually so one failing step
cannot skip the rest.
Fixes#9095
Co-authored-by: Claude Opus 4.8 (1M context) <[email protected]>
Three independent, pre-existing retention bugs kept the classloader of a
finished in-process test run -- and the open jar handles it holds -- alive
for the rest of the server session.
- JUnitXmlTestsListener: testSuite is an InheritableThreadLocal, so
threads spawned during a run (e.g. async-framework pool workers) inherit
a copy of the suite reference that remove() cannot reach.
- TestRecap.collect: the recap is stashed on State.attributes and outlives
the command, so it must not retain live throwables.
- ClassLoaderCache: loaders evicted from delegate by clearExpiredLoaders
are unreachable from the map and never enqueued on the ReferenceQueue, so
neither clear()/close() nor the cleanup thread could ever close them.
SuiteResult now documents the retention hazard on throwables. Adds
deterministic, cross-platform tests for all three severed chains.
The thin client (sbtn) parsed the launcher value flags (-java-home, -mem,
-jvm-debug, -sbt-dir, ...) but then dropped them. The space form was consumed
and discarded; the flag=value form fell through to the residual arguments and
was forwarded to the server verbatim, where --java-home=/path was rejected as a
command (`Not a valid command: --`).
When the client had to start a server (none running, no --server), the consumed
-java-home never reached the forked sbt launcher, so the server came up under
the default JVM and, in CI where the intended JDK is only reachable via
-java-home, failed to connect.
parseArgs now captures the consumed launcher value flags (both `flag value` and
`flag=value`) into Arguments.launcherValueArgs, and the cold-start fork re-passes
them to the sbt launcher so the server runs under the requested JVM. The client
tokenizes arguments by splitting on whitespace, which would otherwise fragment a
value that contains spaces (a Windows path like C:\Program Files\Java); parseArgs
tracks those split boundaries and rejoins a value flag's value. An empty flag=
value and a dangling flag with no value are consumed but not propagated, since
the launcher's require_arg would otherwise fail the fork.
The fork command construction is extracted into a pure, package-visible
serverCommand so a test can assert the propagated flag reaches the started
server. The sbt-launch-jar path is unchanged: it invokes java directly, with no
launcher to interpret the flag.
Fixes#9418
Co-authored-by: Claude Opus 4.8 (1M context) <[email protected]>
Any IOException while creating the boot io socket was wrapped in
ServerAlreadyBootingException and reported as "sbt thinks that server
is already booting" with a stack trace, and non-interactive
invocations exited with code 2. Permission or path-length problems
with XDG_RUNTIME_DIR or the temp directory and Windows named-pipe
access errors all hit this, blocking sbt entirely (#6777). Raw
IOExceptions from the constructor (socket directory creation) were
not caught at all and crashed startup.
getSocketOrExit now connects to the socket (BootServerSocketProbe,
shared with the test suite) to check for a live server before
believing the exception.
Co-authored-by: Claude Fable 5 <[email protected]>
**Problem**
In sbt 2.x, if we execute a run task from the shell,
and if the program fails, it ends up taking down the entire shell
because client-side run rethrows, which is not desirable
for the sbt shell.
**Solution**
1. Omit printing out success for ClientJobParams, which is the runinfo.
2. Print out success or error on the client-side for shell usage case.
The fallback ServerHandler silently dropped requests with unknown
methods, violating the JSON-RPC spec. Clients like Metals that send
unrecognized requests would never receive a response, potentially
causing timeouts or hangs.
Return a -32601 (Method not found) error response instead.
---------
Co-authored-by: Claude Opus 4.6 (1M context) <[email protected]>
* [2.x] feat: Add cacheVersion setting for global cache invalidation
**Problem**
There was no escape hatch to invalidate all task caches when needed.
**Solution**
Add `Global / cacheVersion` setting that incorporates into the cache key
hash. Changing it invalidates all caches. Defaults to reading system
property `sbt.cacheversion`, or else 0L. When 0L, the hash is identical
to the previous behavior (backward compatible).
Fixes#8992
* [2.x] refactor: Simplify BuildWideCacheConfiguration and add cacheVersion test
- Replace auxiliary constructors with default parameter values
- Add unit test verifying cacheVersion invalidates the cache
* [2.x] fix: Restore auxiliary constructors for binary compatibility
* [2.x] test: Improve cacheVersion scripted test and add release note
- Scripted test now verifies cache invalidation via a counter
that increments only when the task body actually executes
- Add release note documenting the cacheVersion setting
**Problem**
sbt bspConfig writes the absolute path of the current Java binary into .bsp/sbt.json. When the user switches Java versions (via sdkman, cs java, etc.) or removes that JDK, the IDE fails to start the sbt BSP server because the hardcoded path is stale or gone.
**Solution**
When an sbt launcher script is available (via `sbt.script` system property or PATH lookup), generate:
"argv": ["/path/to/sbt", "bsp"]
**Problem**
When sbtn (native thin client) can't find a running sbt server, it prompts
to start one. There was no way to opt out of server auto-start from the
client side, which is needed for CI environments and scripting (sbt/sbt#7079).
**Solution**
Reuse the existing sbt.server.autostart system property in NetworkClient
to control whether sbtn should attempt to start a server. When no server is
running and sbt.server.autostart=false, sbtn exits with an error instead
of prompting.
Support setting this property via:
- sbtn --no-server compile (sets sbt.server.autostart=false)
- sbtn --autostart=false compile (new flag following sbt conventions)
- sbtn -Dsbt.server.autostart=false compile (direct system property)
- sbt --autostart=false compile (bash runner and sbtw)
This follows sbt's naming conventions for properties/options: positive
property names with =false opt-out (like --color=false, --supershell=false).
Fixessbt/sbt#7079
Co-authored-by: Claude Opus 4.6 <[email protected]>
The bash launcher's runNativeClient() passed all original CLI args to
sbtn, only stripping --client. This caused sbtn to receive launcher
flags like -mem 10000, which NetworkClient.parseArgs() misclassified:
-mem went to sbtArguments and 10000 went to commandArgs, resulting in
a broken server start command.
**Problem**
When the sbt server fails to start (e.g. wrong JDK version), the client
only shows "failed to connect to server" hiding the actual error. The
server stderr is redirected to /dev/null to prevent Linux pipe buffer
deadlocks (#8442), so diagnostic output is lost.
**Solution**
Redirect server stderr to a temp file instead of /dev/null. When the
server fails to start (portfile never appears), read and print the temp
file contents before throwing ServerFailedException. The temp file is
cleaned up eagerly on both success and failure paths.
Fixes#8812
Closes#4399
- Add subscribeToAll to InitializeOption (protocol); default true for backward compatibility.
- CommandExchange: send broadcast notifyEvent/logMessage only to channels with subscribeToAll.
- TestServer: support subscribeToAll parameter for tests; AbstractServerTest: subscribeToAllForTest.
- ClientSubscriptionTest: assert default client receives build/logMessage when command runs.
- Scripted test server/client-subscription: run show name to exercise server client path.
**Problem**
Forked console currently pulls in full Zinc, which includes JLine.
**Solution**
This implements a lighter-weight, full Java ForkConsoleMain,
which no longer depends on JLine.
Problem
When a user defines an alias with a name that matches an existing task or setting key (e.g., `alias c = compile` when a custom task `c` exists), the alias silently wins and shadows the task.
Solution
Detect conflicts at alias creation time and fail with an error message:
```
Alias 'c' conflicts with a task or setting key of the same name. Use a different alias name to avoid ambiguity.
```
Migrate the following test files from ScalaTest's AnyFlatSpec to
verify.BasicTestSuite, following the pattern established by other
test files in the sbt codebase:
- SizeParserSpec.scala (util-complete)
- MultiParserSpec.scala (main-command)
Changes in all files:
- Replace AnyFlatSpec class with BasicTestSuite object
- Convert 'should ... in' syntax to 'test(...)' syntax
- Use Scala 3 syntax with colon indentation
- Add 'end' markers
- Add explicit types where needed
Related to the ongoing test migration effort.
Fixes#8356
**Problem**
When `sbtn` sends a command while another long-running task (like `console`) is already executing, the client silently blocks with no indication that the command is waiting in a queue.
**Solution**
When a new command arrives via the network channel and another command is currently running, the server now sends an `ExecStatusEvent` notification with status `"Queued"` to the client. The client displays a message like:
```
[info] waiting for: console
```
Fixes#8538
The code was calling e.getMessage.contains() without checking if getMessage()
returns null. Changed to use Option(e.getMessage).exists() to safely handle
null values.
** Problem **
The sbtn (sbt thin client) native image on Linux currently depends on glibc because ipcsocket uses JNI for Unix domain sockets. When building with musl for static linking, the JNI library fails to load since musl doesn't support `dlopen`.
** Solution **
Instead of upgrading to ipcsocket 2.x (which isn't ready for production), I created a `UnixDomainSocketFactory` that detects JDK 17+ at runtime and uses the native `java.net.UnixDomainSocketAddress` API directly via reflection. This completely bypasses JNI on modern JDKs.
For older JDKs (8 and 11), the factory falls back to ipcsocket 1.6.3, which is stable and well-tested.
** How It Works **
The factory checks at startup whether `java.net.UnixDomainSocketAddress` is available:
- **JDK 17+**: Uses native NIO Unix domain sockets (no JNI, no native libraries)
- **JDK 8/11**: Falls back to ipcsocket's JNI-based implementation
This approach:
- Enables musl static linking on JDK 17+ without any native dependencies
- Maintains full backward compatibility with older JDKs
- Keeps the stable ipcsocket 1.6.3 instead of the unstable 2.x
Set window title to 'sbt <command>: <org> % <name> % <version>' when
running sbt run, runMain, bgRun, or bgRunMain.
For server-side runs, window title is set directly. For client-side runs (sbtn), window title is passed via RunInfo protocol
and set by NetworkClient.
Fixes#7586
Previously, some additional code was used to parse the arguments to the
`help` command and ensure that the remaining commands were consistent
with the desired state of sbt.
While this code worked well in sbt 1.x, and was helpful in getting `sbt
help new` to work as expected, this does not work in sbt 2, and is
actually not required.
This patch removes the additional code, which fixes the `help` command
in sbt.
**Problem**
sha256 is currently a bottleneck for no-op compilation.
**Solution**
This adds a local, in-memory cache of sha256 hashes of binary files
using their timestamp and file size.
The size of the digest cache can be configured using localDigestCacheByteSize,
which is set to 1MB by default.
eugene yokota
BrianHotopp
kenji yoshida
Kevin Lee
Anatolii Kmetiuk
bitloi
Dream
Daniil Sivak
BitToby
chrisrock1124
Eve
E.G
aka James4u
MkDev11
Match
calm329
Martin Duhem