[R-pkg-devel] Handling Not-Always-Needed Dependencies?

[Dirk Eddelbuettel]

On 3 August 2016 at 17:35, Duncan Murdoch wrote:
| No, you are reading something into my messages that isn't there.  I 
| think I understand your suggestion now (CRAN should test with no 
| Suggested packages present, not with all of them present).  That is less 
| work than they currently do.

And have Suggests:'ed package tested conditionally.  Whether their tests
execute will be in the control of the one running the tests because he can
make use the optionality provided by Suggests: Either install, or don't.

So it can be the same tests as now.  But better. Because "choice".
 
| I disagree with your suggestion for two reasons.
| 
| 1. I think it would lead to less effective testing.

Nope. See above.
 
| 2. It would be disruptive during the transition.

How to get from here to there is left as an exercise to the reader.

The mostly-wasteful discussion has left me less than motivated to work on
this as there is just so much misunderstanding and blockage.  We could work
all this out during a hackathon day.  Over email, alas, things don't seem to
work too well.

Dirk

-- 
http://dirk.eddelbuettel.com | @eddelbuettel | edd at debian.org