Am Donnerstag, dem 04.08.2022 um 15:37 +0000 schrieb Tobias Geerinckx- Rice: > > To me use of ‘deprecated-package’ in this case is just to ensure > > that users who run ‘guix upgrade’ will transparently get > > ‘fortune-jkirchartz’. I don’t have a strong opinion though. > > I believe what Liliana meant is that it's worse to 'transparently' > upgrade to a data package that doesn't provide the 'fortune' command > at all (or conversely, an implementation of the command that doesn't > propagate the actual fortune data), than it is to signal to users > through an error message that something's up and they need to stop > and think. > > (Er, that's a lot more words to put in someone's mouth than I > intended to, sorry :-) I'm not saying either option is worse than the other, just that we can't pick both in a manner that provides a functioning package, and that transparently upgrading to a package that does nothing is a pretty bad option. We could transparently upgrade to daikichi, but that'd just say "no fortunes found" because FORTUNE_PATH will be empty. Other options would include a metapackage (also rejected by Maxime) or providing a dummy fortune-mod package that fails at build with a helpful message and is disabled in CI (came up as a shower thought).
Cheers
