Skip to content

lkql_jit

Directory actions

More options

Directory actions

More options

Latest commit

 

History

History

lkql_jit

README.md

LKQL JIT

The JIT compiler for the LKQL language.

Dependencies

How to build and run

First, you need to ensure that the mvn command is accessible and fairly recent (at least 3.8.5). Then, make sure you have an accessible Python3 installation, version 3.9 or later. Also make sure that your GraalVM installation is in the 24.2.1 version. Finally, make sure that native-image tool is available in your environment.

1) Install Langkit and make sure it is usable

$[langkit]> pip3 install -r requirement-pypi.txt
$[langkit]> ./manage.py make
$[langkit]> eval `./manage.py printenv`
$[langkit]> export PATH="/path/to/langkit/scripts:$PATH"

If you get a Python error about the Langkit library, add the Langkit directory to your PYTHONPATH environment variable.

Note: If for some reason you are not running ./manage.py printenv, make sure that the sigsegv_handler library is reachable from your LD_LIBRARY_PATH.

Install the Java LangkitSupport library:

$[langkit]> mvn -f langkit/java_support install

2) Build and install Libadalang

$[libadalang]> lkm make --enable-java
$[libadalang]> eval `lkm printenv`

Make sure the Java bindings are installed locally

$[libadalang]> mvn -f build/java install

3) Build and install Liblkqllang

$[langkit-query-language/lkql]> lkm make --enable-java
$[langkit-query-language/lkql]> eval `lkm printenv`

As for Libadalang, make sure Java bindings are installed

$[langkit-query-language/lkql]> mvn -f build/java install

4) Export the needed environment variables :

  • JAVA_HOME should points to your local GraalVM installation
  • GRAAL_HOME should also points to your GraalVM installation

5) Build and run LKQL_JIT

$[langkit-query-language/lkql_jit]> mvn clean package

You can now access LKQL JIT by running the langkit-query-language/lkql_jit/standalone/target/lkql.py script.

This script will launch LKQL JIT with the current accessible Java installation (java -jar ...) so you have to run it in a valid environment.

If you want to build LKQL JIT as a native executable, you can use the native Maven profile:

$[langkit-query-language/lkql_jit]> mvn clean package -P native

Then the lkql executable will be available under the langkit-query-language/lkql_jit/standalone/target directory.

6) Running the code formatter

LKQL JIT project uses prettier-java as code formatter, and the spotless Maven plugin is configured to run it on demand. You can run the following command to format all Java sources:

$[langkit-query-language/lkql_jit]> mvn spotless:apply

However, running it is very prone to error, thus you can install pre-commit hooks to perform this formatting operation on each commit:

$[langkit-query-language]> pre-commit install

Debugging LKQL JIT

To debug LKQL JIT native-image builds you can use the debug maven profile:

$[langkit-query-language/lkql_jit]> mvn clean package -P native,debug

If you need to debug or profile LKQL JIT native-image binaries there is the dev profile which enable support for tools like valgrind, gdb or perf on the produced binaries.

$[langkit-query-language/lkql_jit]> mvn clean package -P native,dev

If you don't need or want any of those, use the prod profile to disable all debugging and profiling information.

$[langkit-query-language/lkql_jit]> mvn clean package -P native,prod