Hello,
Saw your post on the r-devel mailing list, and decided to give a short inquiry/reply here.
I suspect the current proposal you are making, if implemented in R itself, would break backwards-compatibility.
i.e.
t.test(iris[,1,drop=FALSE]) # runs the t-test on the first (and only) column of the data.frame
t.test(iris[,-5]) # runs the t-test on all the numbers in the entire data.frame.
Now your proposal would make these return an error:
library(pipedreams)
t.test(iris[,1,drop = FALSE]) # argument "formula" is missing, with no default
Hence, after this change, if implemented into R core - every code that ran tests this way would begin producing errors.
Can't speak for R core, of course, but I imagine this alone would make the proposed changes unacceptable...
Hello,
Saw your post on the r-devel mailing list, and decided to give a short inquiry/reply here.
I suspect the current proposal you are making, if implemented in R itself, would break backwards-compatibility.
i.e.
Now your proposal would make these return an error:
Hence, after this change, if implemented into R core - every code that ran tests this way would begin producing errors.
Can't speak for R core, of course, but I imagine this alone would make the proposed changes unacceptable...