[R-package] small simplifications in Booster and Predictor

[jameslamb]

Working through #4857, I noticed a few functions in the R package doing something like this:

some_function <- function(x = NULL) {
    if (is.null(x)) {
        x <- -1L
    }
}

This PR proposes simplifying that code by just changing the default values in function signatures.

some_function <- function(x = -1L){
}

Notes for Reviewers

Function defaults are lazily evaluated in R, so no memory is saved by using NULL instead of an integer.

# this does not fail until the function is called
do_something <- function(m = stop("hey")){
     return TRUE
}

Function defaults are not stored in the global environment, so it isn't necessary to worry about side effects like you have to with using mutable types as function defaults in Python.

do_something <- function(x = 10L) {
    x <- x + 1L
    print(x)
}

do_something()
# 11

do_something()
# 11

do_something()
# 11

[StrikerRUS]

LGTM!
Are these functions in public API? If so, this PR should be marked as breaking because it drops possibility of using NULL at a user side.

[jameslamb]

Are these functions in public API? If so, this PR should be marked as breaking because it drops possibility of using NULL at a user side.

Ah good point! I was so focused on this being the mechanism for the default, I didn't think about users actually passing NULL explicitly.

I do consider these methods public, see #4685 (comment) for an explanation.

Given that, I don't support these changes any more 😬 . I don't think this very small simplification is worth breaking users' code. I'm going to close this.

[StrikerRUS]

OK, no problem! 🙂

[github-actions]

This pull request has been automatically locked since there has not been any recent activity since it was closed. To start a new related discussion, open a new issue at https://github.com/microsoft/LightGBM/issues including a reference to this.