CLI vs IDE
We have tried to make every example, tutorial and sample usable via both the command line and the IntelliJ IDE. Most developers will find writing, editing and debugging code more easy with tools such as an IDE. But when a production node is deployed, it will be controlled via the command line - no organisation allows their systems to be running via a developer environment.
IDE - IntelliJ
IntelliJ (R3’s preferred IDE) integrates well with gradle (our chosen build, deployment and CLI tool). IntelliJ understands gradle tasks and dependencies, automatically loading them in the background when a project is first opened or the gradle project changes. Occasionally, however, you may need to refresh the gradle project manually - but this is hinted to you by the IDE. It’s a good idea to do this before carrying on with other work (and in fact you may find it is essential to pick up new libraries, etc.).
There are some great resources about how to get started using IntelliJ. As opposed to trying to repeat them here, we advise you to go to the IntelliJ docs here.
Windows vs Mac / Unix
Due to the nature of their respective command interfaces, gradle is typically ran in windows with the command
gradlew.bat if using the wrapper) and in Mac / Unix environments it is ran via
./gradlew. For brevity, the
simple windows syntax
gradle is used for the majority of the documentation.
As well as including the most significant run and build configurations in the IDE, we also provide gradle tasks to build, install
and run significant parts of Corda demos and tools. Gradle is highly extensible and we use it for downloading required resources,
building components, installing those built components into shared areas, configuring the scripts that run nodes, starting
up demonstration API calls amongst other things. It is exceptionally good at deriving dependency maps and therefore performing
the preceding tasks required in order to do the requested task. However, when confusing build errors manifest, then sometimes
gradle clean may be required in order to clear out any build areas that have an inconsistent state. The total build time
from downloading / cloning the repo to a complete build should be only a few minutes, obviously slightly longer if the
unit tests are run.
Frequently Used Gradle Tasks
Note that the list of tasks can be ran for any gradle project can be displayed by running the task
tasks. Also, note that
gradle is hierarchical and therefore tasks in child directories can be run using a colon separator. For example, if you want to run
the sample attachment demo run configuration
runSender, you would use the command
The most frequent gradle tasks you will probably be running are
build command also executes the
unit tests as well. If you want to build without this level of verification, then use the
assemble command - but we do
not recommend this. After running build, the
install tasks copies over the built jars into the local maven repository
which will then make these available for either the sample code or use with the CorDapp template.
Tasks and processes that are run directly via the IDE (including via the usage of the
driver DSL) can be remotely debugged.
We do not have java debugging currently enabled in the
runnodes scripts generated by a process we refer to as ‘cordformation’
but we will be implementing that shortly.
Via the IDE
To debug: From the IDE, configure the debug connectivity option by the “Edit Configurations” and choosing “+” and then “Remote”.
The debug port start at 5005 and increments for each additional node that starts, the order given by the list in the main
driver configuration (which is primarily listed in the
main function of
Main.kt for each sample. Look for the string
Listening for transport dt_socket at address:5xxx in the log output to determine the exact port for that node. If the log
messages are mixed from several nodes to the same console, then (as earlier stated), the port numbers increment in the order
they are listed in the driver DSL configuration.