dev_help now requires non-installed package

[klmr]

As of release 1.5, dev_help (and ?) on R 3.0.2 (OS X 10.9.2) show the following error:

Error in loadNamespace(name) : there is no package called ‘rstudioapi’

Installing the package manually fixes the problem – but I’m actually not using R Studio, so maybe this package shouldn’t be required to begin with (that said, I don’t see the harm of requiring the package either).

[hadley]

The whole point of rstudioapi is that you don't need rstudio installed.

@wch I think this means that rstudioapi should be in imports, not suggest.

[wch]

@hadley We could put the rstudioapi in imports, but I think it shouldn't be hard to structure the code to avoid attempts to load rstudioapi.

@klmr Can you include a traceback for the error? I'm curious to see where it's trying to load the package.

[hadley]

I'm pretty sure rstudioapi belongs in imports, because the goal of the package is abstract away code which checks if rstudio is running. Adding extra code to check if rstudioapi is installed sort of defeats the purpose. (The check is probably coming from inside dev_help()

[klmr]

@wch Here you go:

> traceback()
11: stop(e)
10: value[[3L]](cond)
9: tryCatchOne(expr, names, parentenv, handlers[[1L]])
8: tryCatchList(expr, classes, parentenv, handlers)
7: tryCatch(loadNamespace(name), error = function(e) stop(e))
6: getNamespace(ns)
5: asNamespace(ns)
4: getExportedValue(pkg, name)
3: rstudioapi::hasFun
2: dev_help(e1_str)
1: `?`(test)

[wch]

@hadley OK, that makes sense.

@klmr Thanks, my version was slightly outdated so I didn't have that code path.