Open waynelapierre opened 1 year ago
I'm sure if Microsoft, or any other company, wants to set a team to make R faster without increasing the maintenance burden to the R core team it will be welcomed. I'm not speaking by the R core but I'm sure there will be interest of other people if you want to contribute to make it faster. Maybe the R consortium would be interested in funding and setting a working group together. In my experience, ALTREP and other approaches have made R faster.
Not sure if pqR is still maintained, but its developer Radford Neal seems to have been working on making R faster for a long time. https://github.com/radfordneal/pqR
@waynelapierre some changes from pqR have been incorporated into R, so far as I know, but many others required too-drastic changes to the internals and/or interfered with things R-core members were working on. Also, from the looks of that repository, no, it does not appear to be currently maintained as there haven't been any commits to it in ~3 years.
Occasionally there are suggestions for improving performance on the R-devel mailing list from volunteers, but they rarely get response. Here I have listed some: https://github.com/HenrikBengtsson/Wishlist-for-R/issues/141. In particular, the improvements proposed by Sebastian Krantz (based on kit
and collapse
packages) look interesting.
Python has a team from Microsoft that is making Python faster: https://www.phoronix.com/review/python-311-performance