Should `skip_if_offline()` imply `skip_on_cran()`? Or at least default to that?

[jennybc]

I recently learned the hard way that a test for which I have skip_if_offline() can still hit a secondary GitHub API rate limit on CRAN, causing intermittent test failure. Clearly this test also needs skip_on_cran().

This made me think: maybe that is generally true? Or almost always true?

I think it's worth considering if skip_if_offline() should also skip on CRAN or have an argument to control this, defaulting to TRUE.

[hadley]

I guess I never actually use skip_if_offline(); I just use skip_on_cran(). So I don't have any opinion either way.

[jennybc]

I think I use them individually in order to be self-documenting and super-clear about intent. But I can't think of a single test that hits the internet that I would want to run on CRAN. I rely on the test data I am getting constantly from GHA for such tests.