Add envir_prep to checker function call

[schloerke]

Fixes #163

Submits a copy of the exercise environment before the submission is rendered. This gives a fresh environment for the checker to user that contains the setup code.

[jjallaire]

We try not to have so many spurious whitespace changes in a diff (makes it hard to tell what material code changes there are). It would be great if you could remove these.

If you feel strongly about enforcing a whitespace convention I'd suggest just one big commit that fixes all the whitespace (and updates the .Rproj file to have a setting that agrees with the scheme).

[schloerke]

@jjallaire Thanks for the feedback! I've reduced the changes. (I'll make a separate PR for .Rproj whitespace changes later.)

I dropped compiling within the ./docs folder as it produces many changes. Similar to shiny, I think we can wait until CRAN release.

• NEWS item
• test?

[jjallaire]

Great, thanks!

[wch]

Should this use all.names=TRUE?

[wch]

Some thoughs that aren't super important, but they might be slightly useful because the function is exported....

It might be faster with this, though you should probably test.

list2env(as.list.environment(envir, all.names = TRUE, sorted = FALSE))

Some notes:

The sorted=FALSE is because sorting can take a nontrivial amount of time.

system.time(ls(baseenv()))
#>   user  system elapsed 
#>  0.007   0.000   0.006 

system.time(ls(baseenv(), sorted=FALSE))
#>   user  system elapsed 
#>  0.003   0.000   0.003 

The reason it uses an explicit as.list.environment is because if the environment happens to have a class on it, it will error. For example:

e <- new.env()
e$x <- 1
class(e) <- "foo"

as.list(e)
#> Error in as.vector(x, "list") : 
#>   cannot coerce type 'environment' to vector of type 'list'

as.list.environment(e)
#> $x
#> [1] 1

[wch]

Change to duplicate_env, and export this.