I'm going to give it another try now that I have some time.
Making everyone who ever changes base have to go through a public committee and do an entire analysis of the impact on hackage is not a good way to have new contributors to base. Maybe that is what is wanted for something as delicate as base but I think you should acknowledge that the standards are extremely exacting and offputting. If @trac-isovector doesn't have any hope what is left for the rest of us?
I just ran into an edge case where this would be very useful trying to encode Structure of Array style arrays using generics and data families (essentially identical to how Data.Vector.Unboxed
works). Are we any closer to this issue being solved?
I don't think it should be supported but maybe worth having an error message for it?
When I run ghcup set
it doesn't seem to be able to find the versions of GHC installed. For example I have the following:
> ghcup install ghc-8.10.4
[ Warn ] This is an old-style command for installing GHC. Use 'ghcup install ghc' instead.
[ Warn ] GHC ver 8.10.4 already installed; if you really want to reinstall it, you may want to run 'ghcup rm ghc 8.10.4' first
> ghcup set ghc-8.10.4
[ Warn ] This is an old-style command for setting GHC. Use 'ghcup set ghc' instead.
[ Error ] The version "ghc-8.10.4" of the tool "GHC" is not installed.
Any ideas what might be causing this? I also find the warnings very confusing since the commands shown are also on the gitlab readme.