I've just run into an old R script of mine from July 2013, whose goal was to create a map. It is written in a rather inefficient way, but what is amazing about it is that it still runs and can still create the very same map! #RStats
1
5
32
That’s rare without docker. I couldn’t even run a code I wrote 2 years ago b/c some tidyverse functions got deprecated…
1
2
Replying to @HammadSheikh @signormirko @rstatstweet
See the tweet below -- it can work very well and eg ​I (very publically via cron) run #RStats code from 14 years ago (CRANberries) every hour and from 8 years ago (CRAN Repo Policy bot) daily.
Dirk Eddelbuettel @eddelbuettel
26 Sep 2021
"The biggest piece is minimizing your dependencies, and limiting them to ones that value backwards compatibility." While primarily about 'low-upkeep' software, this generalises to #rstats in #production, to #reproducibility and to good #science. jefftk.com/p/designing-low-u…

Sep 26, 2021 · 8:23 PM UTC

2
1
2
Replying to @eddelbuettel @HammadSheikh @rstatstweet
I agree with you, @eddelbuettel on minimizing dependencies. I find it a very good principle in general, both for my own analyses and for the packages I develop.(By the way, the old script relied only on the #rworldmap package)
2
Replying to @eddelbuettel @HammadSheikh @signormirko @rstatstweet
How do you actually achieve this when you are not a developer by trade? (The article provided discusses the general principles but falls short on any concrete examples). Do you count the number of dependencies D for an R package and drop the package if D > what?
2
1