The error is because of a previous use of magic::magic_for which was somehow being brought back into the environment. The apparent intent of the function (no changes since 2016) is to allow a for loop to store the results of a modified function, something like:
library(magicfor) # Load library
magic_for(print) # Call magic_for()
for (i in 1:3) {
squared <- i ^ 2
print(squared)
}
#> The loop is magicalized with print().
#> [1] 1
#> [1] 4
#> [1] 9
magic_result_as_vector() # Get the result
#> [1] 1 4 9
It appears that the "magicalization" of the print function was preserved (perhaps in .Rdata), but on a session restart, the magicfor package was not available (or just not loaded). I'd think that if the magicfor package is truly required for a workflow, then (1) make sure it's loaded, and (2) submit a bug to the author so that it better accommodates this scenario.
Lacking that, my suggestion for now:
- do not load
.Rdata files ("saved session" or similar); many feel it is far better to have a reproducible workflow and regenerate the intermediate/final structures you need vice loading them; the exception to this suggestion would be for long-running processes, but even then I think the better case is to explicitly save/cache just those values, not the entire session;
- similarly, check
.Rprofile and other R startup files to remove any mention of magicfor;
- perhaps nuclear, but a community thread mentioned uninstalling the
magicfor package; it's unclear if that was part of the resolution or not, though it seems unlikely if the underlying cause was a stale reference in an .Rdata file.
