search

nimble-dev / nimble

The base NIMBLE package for R
http://R-nimble.org
BSD 3-Clause "New" or "Revised" License
160 stars 24 forks source link

add formal dependency on R 4.1.0? #1463

Closed paciorek closed 5 months ago

paciorek commented 5 months ago

@perrydv has used |> and \(x), which rely on R >=4.1.0. We formally depend on R 3.1.2 but in practice probably on something newer.

This is just to open a (hopefully quick) discussion on whether to change to 4.1.0.

Looping @danielturek in on this.

danielturek commented 5 months ago

Not on a computer, but lodging a tertiary vote to not upgrade dependency to R 4.x.x. Unless it’s necessary to support our codebase, which I don’t believe in the case, is there any reason to do so?

On Tue, Jun 4, 2024 at 6:16 PM Christopher Paciorek < @.***> wrote:

@perrydv https://github.com/perrydv has used |> and (x), which rely on R >=4.1.0. We formally depend on R 3.1.2 but in practice probably on something newer.

This is just to open a (hopefully quick) discussion on whether to change to 4.1.0.

Looping @danielturek https://github.com/danielturek in on this.

— Reply to this email directly, view it on GitHub https://github.com/nimble-dev/nimble/issues/1463, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABCNNYLUZZ6S6P4JLHUVN5LZFY4D7AVCNFSM6AAAAABIZONVCSVHI2DSMVQWIX3LMV43ASLTON2WKOZSGMZTINBWGE2TKMY . You are receiving this because you were mentioned.Message ID: @.***>

perrydv commented 5 months ago

It may have been cavalier of me, but 4.1.0 was released just >3 years ago, so I'm not sure how reasonable it is to say we support older versions than that in any case?

paciorek commented 5 months ago

I'm wavering here. On one hand, I agree with Daniel. OTOH, given all the headaches with this release, I'm also not feeling like we want to go back in and back-transform the code Perry added that uses this, even though it seems trivial.

paciorek commented 5 months ago

I made the call (with ok from @perrydv ) to revert this and not depend on >= 4.1.0.

What pushed me to that was thinking about a use case of a user using nimble on a multi-user system who doesn't have the ability to unilaterally update R. Not sure if that would happen but we've seen cases where users had limited flexibility of that sort.