jvm
Options for general JVM functionality.
JDK strings will be passed directly to Coursier's --jvm
parameter. Run cs java --available
to see a list of available JVM versions on your platform.
If the string 'system'
is passed, Coursier's --system-jvm
option will be used instead, but note that this can lead to inconsistent behavior since the JVM version will be whatever happens to be found first on the system's PATH.
Backend: pants.backend.experimental.java
Config section: [jvm]
Basic options
debug_args
--jvm-debug-args="['<str>', '<str>', ...]"
PANTS_JVM_DEBUG_ARGS
[jvm]
debug_args = [
'<str>',
'<str>',
...,
]
[]
Extra JVM arguments to use when running tests in debug mode.
For example, if you want to attach a remote debugger, use something like ['-agentlib:jdwp=transport=dt_socket,server=y,suspend=y,address=5005']
.
default_resolve
--jvm-default-resolve=<str>
PANTS_JVM_DEFAULT_RESOLVE
[jvm]
default_resolve = <str>
jvm-default
The default value used for the resolve
and compatible_resolves
fields.
The name must be defined as a resolve in [jvm].resolves
.
deploy_jar_exclude_files
--jvm-deploy-jar-exclude-files="['<str>', '<str>', ...]"
PANTS_JVM_DEPLOY_JAR_EXCLUDE_FILES
[jvm]
deploy_jar_exclude_files = [
'<str>',
'<str>',
...,
]
[ "^META-INF/[^/]+\.SF$", "^META-INF/[^/]+\.DSA$", "^META-INF/[^/]+\.RSA$", "META-INF/INDEX.LIST$" ]
A list of patterns to exclude from all deploy jars. An individual deploy_jar target can also exclude other files, in addition to these, by setting its exclude_files
field.
resolves
--jvm-resolves="{'key1': val1, 'key2': val2, ...}"
PANTS_JVM_RESOLVES
[jvm.resolves]
key1 = val1
key2 = val2
...
{ "jvm-default": "3rdparty/jvm/default.lock" }
A dictionary mapping resolve names to the path of their lockfile.
Advanced options
global_options
--jvm-global-options="['<str>', '<str>', ...]"
PANTS_JVM_GLOBAL_OPTIONS
[jvm]
global_options = [
'<str>',
'<str>',
...,
]
[]
List of JVM options to pass to all JVM processes.
Options set here will be used by any JVM processes required by Pants, with the exception of heap memory settings like -Xmx
, which need to be set using [GLOBAL].process_total_child_memory_usage
and [GLOBAL].process_per_child_memory_usage
.
jvm_global_options
on local_environment
, docker_environment
, or remote_environment
targets.jdk
--jvm-jdk=<str>
PANTS_JVM_JDK
[jvm]
jdk = <str>
temurin:1.11
The JDK to use.
This string will be passed directly to Coursier's --jvm
parameter. Run cs java --available
to see a list of available JVM versions on your platform.
If the string 'system'
is passed, Coursier's --system-jvm
option will be used instead, but note that this can lead to inconsistent behavior since the JVM version will be whatever happens to be found first on the system's PATH.
nailgun_remote_cache_speculation_delay
--jvm-nailgun-remote-cache-speculation-delay=<int>
PANTS_JVM_NAILGUN_REMOTE_CACHE_SPECULATION_DELAY
[jvm]
nailgun_remote_cache_speculation_delay = <int>
1000
The time in milliseconds to delay speculation of nailgun processes while reading from the remote cache.
When speculating, a remote cache hit will cancel the local copy of a process. But because nailgun does not natively support cancellation, that requires killing a nailgun server, which will mean that future processes take longer to warm up.
This setting allows for trading off waiting for potentially slow cache entries against potentially having to warm up a new nailgun server.
reproducible_jars
--[no-]jvm-reproducible-jars
PANTS_JVM_REPRODUCIBLE_JARS
[jvm]
reproducible_jars = <bool>
False
When enabled, JAR files produced by JVM tools will have timestamps stripped.
Because some compilers do not support this step as a native operation, it can have a performance cost, and is not enabled by default.
tool_jdk
--jvm-tool-jdk=<str>
PANTS_JVM_TOOL_JDK
[jvm]
tool_jdk = <str>
temurin:1.11
The JDK to use when building and running Pants' internal JVM support code and other non-compiler tools. See jvm
help for supported values.
Deprecated options
None
Related subsystems
None