Examples generate errors

[arilamstein]

One thing that R CMD check does is sequentially run all the examples in a package. If a single example fails, it reports the error and stops testing the examples. This is a useful test, and (I think) a reason to make sure that our examples do not generate errors.

However, it looks like the last example for parse_integers is perhaps designed to fail. I handled this by wrapping the whole set of examples in a \dontrun block. But then it appears that the next function (str_relevel) has the same issue: the last example throws an error. I can't be sure, but I think that this one is intentional as well.

At this point I decided to create this issue instead of wrapping the examples in a \dontrun or removing the last one. While I do think that having R CMD check run all the examples and report on errors is useful, it's possible you don't agree (or just weren't using / aware of this feature). If you'd like to use it, though, I recommend making sure that no examples generate errors.

[arilamstein]

Quick update: I decided to rollback my change that wrapped the examples in a \dontrun. The reason is that in that PR I specifically said that, in addition to moving scales from Depends: to Imports: there was nothing controversial in the PR. While this isn't exactly controversial, I do think that there's room for discussion about the best approach, so I thought best to leave it out.

Incidentially, I checked and it turns out that you can have comments in examples. So we could change the last example from:

parse_integers("foo")

to:

#' # causes an error because "foo" is not an integer.
#' # parse_integers("foo")