stylistic changes

[IndrajeetPatil]

No description provided.

[strengejacke]

sadly, this option doesn't seem to work on Windows...

library(parameters)
model <- lm(mpg ~ wt + cyl, data = mtcars)

microbenchmark::microbenchmark(
  model_parameters(model, bootstrap = TRUE, iterations = 1000, parallel = "snow"),
  times = 5
)
#> Unit: seconds
#>                                                                                  expr
#>  model_parameters(model, bootstrap = TRUE, iterations = 1000,      parallel = "snow")
#>       min       lq     mean   median       uq      max neval
#>  1.072621 1.076588 1.141162 1.108037 1.109053 1.339513     5

microbenchmark::microbenchmark(
  model_parameters(model, bootstrap = TRUE, iterations = 1000, parallel = "no"),
  times = 5
)
#> Unit: seconds
#>                                                                                expr
#>  model_parameters(model, bootstrap = TRUE, iterations = 1000,      parallel = "no")
#>       min       lq     mean   median       uq      max neval
#>  1.050956 1.060096 1.118408 1.064866 1.070488 1.345632     5

microbenchmark::microbenchmark(
  model_parameters(model, bootstrap = TRUE, iterations = 1000, parallel = "multicore"),
  times = 5
)
#> Unit: seconds
#>                                                                                       expr
#>  model_parameters(model, bootstrap = TRUE, iterations = 1000,      parallel = "multicore")
#>       min       lq     mean   median       uq      max neval
#>  1.083162 1.138866 1.161232 1.176707 1.188325 1.219101     5

^{Created on 2021-03-09 by the reprex package (v1.0.0)}

[IndrajeetPatil]

I will merge for now, and we can build on it further later.

[strengejacke]

But checks don't pass.

[IndrajeetPatil]

Uh-oh, let me check what happened.

[strengejacke]

Looks like an invalid argument was used in the tests, which didn't do any harm, because we didn't pass down the dots to boot() ;-)
When we use boostrapping, dots may only contain valid arguments for boot().

[strengejacke]

@DominiqueMakowski This PR breaks a lot of other stuff, in particular all parameters that are passed down to bayestestR::describe_posterior(). Not sure what's the best way to go... Maybe we can capture the dots, then "filter" for valid arguments for boot(), and only pass those valid arguments to boot() (via do.call()).

@IndrajeetPatil would you revise this PR accordingly to not break current functionality?

[DominiqueMakowski]

Maybe we can capture the dots, then "filter" for valid arguments

wouldn't something like parallel <- ifelse("parallel" %in% names(as.list(...)), as.list(...)$parallel, "no") and then pass parallel to boot work?

[IndrajeetPatil]

I prefer to then just go back to the earlier implementation I had in mind:
explicitly providing parallel and ncpus argument, which, yes, complicates the API slightly, but would not cause any problems for any of the methods.

[IndrajeetPatil]

As discussed in #436, I've decided to revert all boot-related changes. Now the PR only contains stylistic 80-width changes.