refactor of initial plugin

[jagedn]

• upgrade to nextflow 24
• use carpet lib
• refactor splitParquet
• some validation use cases

This is an initial upgrade of the plugin using an open source library (carpet) to read/write parquet files avoiding the complexity of using low level library.

Carpet can read parquet files as Map but also you can provide a java Record class as projection so the library read only field presents in the Record

This PR contains a validation folder, with a few small parquet files, to validate the plugin (tests will be added in next PRs) so you can see in action following these steps:

./gradlew copyPluginZip
cd validation
NXF_PLUGINS_DIR=$(pwd)/../build/plugins nextflow run read_raw.nf 
NXF_PLUGINS_DIR=$(pwd)/../build/plugins nextflow run read.nf 
NXF_PLUGINS_DIR=$(pwd)/../build/plugins nextflow run read_complex.nf 

[pditommaso]

Great Jorge, the README is still valid?

[jagedn]

Great Jorge, the README is still valid?

Yes, the method names are the same (although write method is empty by the moment)

[pditommaso]

Awesome!

[jerolba]

@jagedn FYI: just yesterday I released version 0.2.0:
https://mvnrepository.com/artifact/com.jerolba/carpet-record/0.2.0
Relevant changes :

• upgrade to Parquet 1.14.2
• review transistive dependencies exclusions if you are not using hadoop (reduces 2.5MB)