Message-ID: <CABtg=Kn2S344dW2gzeLmxbsbi0vjJ+r7z1qmcKkVryN7WYQXtQ@mail.gmail.com>
Date: 2016-08-04T16:56:22Z
From: Gábor Csárdi
Subject: [R-pkg-devel] Handling Not-Always-Needed Dependencies? - Part 2
In-Reply-To: <5e698b95-744b-dfd7-217e-35fa558c7b2d@gmail.com>
On Thu, Aug 4, 2016 at 5:48 PM, Duncan Murdoch <murdoch.duncan at gmail.com> wrote:
[...]
> I'd say it's up to you as the author of the test. Would skipping that test
> mean that your package was not adequately tested? If so, then you should
> get an error if it isn't available, because otherwise people will think
> they've done adequate testing when they haven't. One way this could happen
> if a major function of your package is being tested on a sample dataset
> from a Suggested package. Users of your package don't need the other one,
> but testers do.
Indeed. IMO this was anything but a yes/no question. :)
Personally I would love to be able to have BuildImports and
TestImports. The former for build time dependencies, the latter for
dependencies when running the tests.
But I understand that the current system is complicated enough.
Gabor
> Duncan Murdoch
>
> ______________________________________________
> R-package-devel at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-package-devel