First draft of scope_()

[krlmlr]

Fixes #28 (eventually).

@jimhester: That was easy.

@kevinushey: Thanks for your package. Try, for instance, scope_(setwd). Any plans for CRAN release yet? ;-)

[jimhester]

@krlmlr you will have to explicitly attach devtools to get Remotes: to work on travis with CRAN devtools. Due to r-lib/devtools#914

[kevinushey]

That's awesome! Some definite wizardry going on in the scope_ definition 😄

I'm still waiting a bit to convince myself that the defer_parent implementation will work as expected in all contexts; I think I will let it 'bake' for a bit longer and submit after that.

[krlmlr]

Thanks. Do you think it's better to add .scope_env = parent.frame() to the formals of the generated function, instead of relying on defer_parent?

[kevinushey]

Sorry, I'm a bit foggy on what you mean -- can you elaborate a bit more?

[krlmlr]

Currently:

> scope_(setwd)
function (dir) 
{
    old <- setwd(dir = dir)
    later::defer_parent(setwd(old))
    old
}

But it could be:

> scope_(setwd)
function (dir, .scope_env = parent.frame()) 
{
    old <- setwd(dir = dir)
    later::defer(setwd(old), envir = .scope_env)
    old
}

[jimhester]

Ended up merging this manually (54becef) and doing the rest of the work to get it all the functions integrated + tests.

It mostly uses this implementation plus @krlmlr's suggestion to use .scoped_envir, I also ended up calling the functions scoped_ as that sounds a little nicer to me.

@kevinushey Were you planning on submitting later to CRAN soon, if so we can coordinate a withr release as well.