Bump up R version requirement

[khufkens]

Increase the mandatory R version requirement to 3.6?

In general I've run into many troubles, either personally or when giving workshops when people run outdated R installs. Pushing for a 3.6 requirement might trap some issues which are not necessarily package related (not in 3.6 anyway).

https://github.com/khufkens/ecmwfr/blob/7ed5c96d969bf5b733671be8313350d03aa54de4/DESCRIPTION#L31

and potentially related to this: #49 (comment)

Thoughts @eliocamp ?

[eliocamp]

I don't know. I wouldn't want to increase the requirements unless we are using some functionality that is not available in previous versions and hard to backport. I am not aware of any used in this package.

I use R 3.6 in all of my machines and in the last workshop I gave I used RStudio Cloud, which I think also has the most up-to-date version of R. So I don't have much experience working with old versions. What has been your experience? Does ecmwfr fail often with R < 3.6?

[khufkens]

You are probably right, problem is that I don't test anymore on 3.4. I used to have at least one system running on 3.4 but I've upgraded all of them to >3.4. Otherwise the issue with wf_request() would have been caught earlier. I'll include a new line in the travis file.

[khufkens]

Fixed with Travis CI integration.

[eliocamp]

For what is worth,we might get away with a lower version if we wanted to.

https://ateucher.shinyapps.io/check_r_versions_of_package_dependencies/

Is 3.3.0 available to test in Travis?

[eliocamp]

Now I'm looking at the issue backwards and trying to figure out if it's possible to bump down the R version required. 🤣

Looking at the dependencies, why is {stars} suggested? I don't see any stars:: or library(stars) anywhere in the code. Is it a vestigial dependency?

[khufkens]

I would keep it at 3.4, low enough. If someone asks advice on <3.4 I would suggest to update first. I used stars in some of the vignettes / docs. It might be deprecated.

…

On Fri, 3 Jan 2020 at 16:15, Elio Campitelli ***@***.***> wrote: Now I'm looking at the issue backwards and trying to figure out if it's possible to bump *down* the R version required. 🤣 Looking at the dependencies, why is {stars} suggested? I don't see any stars:: or library(stars) anywhere in the code. Is it a vestigial dependency? — You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub <#51>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAKKUERTQUCY3CT2VPRAGBDQ35JBFANCNFSM4KCFXRWA> .

-- Koen Hufkens, Ph.D., Marie Skłodowska-Curie Fellow INRA Bordeaux Aquitaine Water Relations and Ecosystem Functioning (EcoFun) team Ghent University Computational & Applied Vegetation Ecology Lab Congo basin eco-climatological data recovery and valorisation (COBECORE) project lead (www.cobecore.org) @koen_hufkens - www.khufkens.com Find the Rhythm of Tropical Trees - www.junglerhythms.org Explore Seasonal Changes in a Virtual Forest - www.virtualforest.io