I pushed some more on this and managed to get something which "works".
That is, I was able to declare a Data instance for T:
data T (phantom :: k) = T
instance (Typeable k, Typeable phantom, Possibly Data phantom) =>
Data (T (phantom :: k)) where
...
dataCast1 = dataCast1T
(Mo
Hi all,
MSYS2 maintainers have agreed that the change by Cygwin constitudes a
regression for the MSYS2 project and patched out the changes.
It's now safe to upgrade MSYS2 again.
Tamar
On Mon, Feb 20, 2017 at 3:58 PM, Phyx wrote:
> Hi Windows devs,
>
> February comes with a major change in beh
Yikes, I was afraid of reaching for something like DataIfStar. But I
don't have any other ideas, so let's play along with that example.
Frustratingly, I can't even write the class definition of DataIfStar.
If I do, GHC yells at me like so:
• Expected a type, but ‘k’ has kind ‘t’
• In the
NB: the actual ticket Trac #12689 is /not/ about rules /for/ data cons. It's
about rules that /match/ datacons. It's only the latter I object to. The test
T12689 sort of snuck in there under false pretences :-).
Also I'm not permanently set against rules for datacons. It's just that I
think
