New Post: An Ode to Stable Interfaces, and Praise To Base R and R Core In which I run code I posted fourteen years ago: it works unaltered. That should be cherished, celebrated--and encouraged. We need more examples like this. dirk.eddelbuettel.com/blog/2… #rstats #reproducibility

Mar 20, 2021 · 8:43 PM UTC

4
58
14
271
Replying to @eddelbuettel
People assume stability means all the blog post of 2014-2016 about #RStats being slow are still true... You should have added a benchmark for both versions as well... 😃😃
1
1
Replying to @eddelbuettel
Totally agree! This reminds me of an interview I saw with a Bjarne Stroustrup that C++ should be stable so that it ran the same code over 30 years in places it was he’d to change in production (I believe the example he gave was oil platforms).
2
Replying to @eddelbuettel
Indeed great. But given the raise of environments for reproducible computing like docker (in particular rocker), conda and renv, would you agree that it’s more ok than in the past to give away some of the API stability for improving APIs in the long term?
3
3
Replying to @eddelbuettel
That's impressive and YES it must be celebrated and aimed as a goal.
5