This is an SLR and LR(1) parser generator with some support for common LR(k) constructs. It can generate outputs for arbitrary target languages. It will use all cores on your build system, and is engineered for maximum practicality and performance.
(TODO)
Run bootstrap.sh
then upgrade.sh
.
Polyglot is published on Maven Central as
org.anarres.polyglot:polyglot-core:${version}
.
The Gradle plugin, Maven plugin and Ant task are published
as polyglot-gradle
, polyglot-maven
and polyglot-ant
respectively. There is also a commandline version, published as
org.anarres.polyglot:polyglot-cmd:${version}
.
Gradle is the Right Answer(tm).
buildscript {
classpath 'org.anarres.polyglot:polyglot-gradle:1.0.0-SNAPSHOT'
}
apply plugin: 'org.anarres.polyglot'
See PolyglotPlugin.
If you understand Maven better than I do, please help to correct this.
<plugin>
<groupId>org.anarres.polyglot</groupId>
<artifactId>polyglot-maven</artifactId>
<version>${polyglot.version}</version>
<executions>
<execution>
<goals>
<goal>polyglot</goal>
</goals>
</execution>
</executions>
</plugin>
You will also need to set up a dependency on
org.anarres.polyglot:polyglot-runtime:${polyglot.version}
.
See PolyglotMojo.
My commiserations on still using ant. The ant task is called
org.anarres.polyglot.ant.Polyglot
You will also need to set up a dependency on
org.anarres.polyglot:polyglot-runtime:${polyglot.version}
.
See PolyglotTask.
Polyglot aims for the maximum possible performance for the practical developer. It will attempt fast strategies (SLR), and if they fail, it will attempt slower strategies (LR(1), LR(k)). It will use all CPU cores on the system, so your computer will go out to lunch during a build. However, the performance for LR(1) universes in excess of 10 million items is quite acceptable.
The polyglot core is also designed to be almost steady state in the garbage collector, although the template engine (currently Velocity) is quite inefficient and should be replaced.
Polyglot is based on the specification (but not the code) of SableCC by Etienne Gagnon, and aims for limited backwards compatibility with all SableCC grammars and applications built on the generated code.