Here are my brief notes from the meeting:
* Alex will make the Renjin package build system available in the coming week or so.
* We will use this mailing list to discuss the progress as well as github issues where appropriate
* The current minimum set of packages consists of 950 packages (all coming from a handful of high level packages such as ggplot2 etc.)
* There is a blocklist for certain packages that will never work in Renjin (e.g. ones that depend on R internals)
* Due to various reasons it is unrealistic to require 100% tests success so some way to determine which test results that can be ignore is needed
* There is already a way to provide a replacement package for a package that will not work in Renjin in the package build system.
* The initial goal would be to have a reliable build, powered by the new gradle scripts, with successful tests working for those 950 "core" packages serving as a comprehensive regression suite.
* Once this is working, Alex et. al. have some very interesting ideas about optimizations, debugging R code etc. that could be enabled
* Other packages outside of the core 950 could probably be added and published to Renjin CRAN etc. but should probably be excluded from the core suite.
* We target R 3.5 for now and later figure out what to do when Renjin compatibility targets R 4.0.x
We were not that many in the call but I felt it was an exciting start of something I really believe will help in attracting more R/Java developers to Renjin and broaden the user base.
Please correct me if there is anything I got wrong.
Best regards,
Per