From the help information:
-J<flag> Pass <flag> directly to the runtime system.
Use one -J for each runtime flag or flag argument
-R<flag> Pass <flag> to the remote runtime system.
Use one -R for each remote flag or flag argument
I cannot find an explanation in both the tools documentation and jshell user guide.
Also, what is "the remote runtime system" in the context of jshell?
As I understand it, JShell has 3 main 'places' to execute code:
In the current process (see DirectExecutionControl)
In the same JVM as the JShell client (see LocalExecutionControl)
On the remote agent (see JdiDefaultExecutionControl)
Using jshell
tool, we have no current process before launch, so we have only two options - use one JVM (locally), or use two JVMs - one for JShell client (locally) and another for the execution engine (possibly remotely).
The interesting thing is, JShell always launch two JVMs by default, as the hard-coded --execution key is "failover:0(jdi:hostname(" + loopback + ")),1(jdi:launch(true)), 2(jdi)"
(see JShell class source code).
Closer to the point. I've made a couple of experiments with -verbose
option and checked JVM options in runtime with ManagementFactory.getRuntimeMXBean().getInputArguments()
.
jshell -J-verbose
commandPrinted
-verbose
output in the console.No
-verbose
option in the input arguments:[-agentlib:jdwp=transport=dt_socket,address=127.0.0.1:63305]
jshell -R-verbose
commandNo
-verbose
output in the console.Printed
-verbose
option in the input arguments:[-agentlib:jdwp=transport=dt_socket,address=127.0.0.1:63339, -verbose]
jshell --execution="local" -J-verbose
commandPrinted
-verbose
output in the console.Printed
-verbose
option in the input arguments:[-Dapplication.home=C:\Program Files\Java\jdk-9, -Xms8m, -verbose, -Djdk.module.main=jdk.jshell]
jshell --execution="local" -R-verbose
No
-verbose
output in the console.No
-verbose
option in the input arguments:[-Dapplication.home=C:\Program Files\Java\jdk-9, -Xms8m, -Djdk.module.main=jdk.jshell]
TL;DR
Remote execution (default case, execution over JDI)
-J<flag>
passes option to the JShell client JVM
-R<flag>
passes option to the execution engine JVM
Local execution (--execution="local")
-J<flag>
passes option to the only present JVM
-R<flag>
does nothing
I still find the scope of explaining the usage of the flags used in both the attributes for the question to be answered completely, hence putting that in words here.
➜ -J
flag is used to provide a runtime argument to the JShell which is similar to the way one provides while performing via an IDE under Run -> Configuration to specify arguments as -Dkey=value
.
The usage of the attribute is documented and quite similar to ones illustrated for the flag -C
just that the -J flags are instead the java
command line options.
For example, using the -XX:+PrintCommandLineFlags
would detail out the runtime flags used by the current JVM.
So, the default values of command line flags used by your JShell instance(without setting any additional flag) could be as:-
But let's say you don't want to make use of the CompactStrings
of Java9 in your JShell execution you can tell the JVM using the -J
flag and -XX:-CompactStrings
attribute to do so, as -
jshell -R-XX:+PrintCommandLineFlags -R-XX:-CompactStrings
would list out the following output:
➜ Similar java
command line options/flags, when attached with and used to start JShell on a remote JVM, are linked to -R
attribute of the JShell during remote execution.
jshell -R-XX:+PrintCommandLineFlags -R-XX:-CompactStrings
Anatoly's answer has got some good amount of research attached to it and I would suggest that to be read to understand about control and execution engine of the JShell for local Vs remote execution.
来源:https://stackoverflow.com/questions/46399679/what-is-the-exact-meaning-purpose-of-the-j-and-r-flags-in-jshell