-
Notifications
You must be signed in to change notification settings - Fork 25
Factor out hfunctor and friends into its own package #170
Comments
Thinking about the abstraction behind Originally,
But In quickcheck-state-machine, the user-defined type The variable
This Note that hedgehog's
To make the kinds match, we would need to write
|
Neither choice is very satisfying. I guess the The first release of @jystic: Thoughts? |
I would only say that I don't mind too much if there is extra cruft in the internal usage of the abstraction, but I would prefer that the way clients use the API is as clean as possible. As such the
|
The user still writes almost the same code with We could define a "quickcheck-state-machine"-specific
|
Now that you explain it, both solution sound pretty reasonable, especially if the hard work is hidden behind TH. Do you have a preference? |
I prefer |
(I'm fine with either.) |
I'm ok with |
Another thing worth considering: how much does the GADT give us? Could it be worth scraping it, so we can use |
I just found that the rank2classes has the higher-order |
We ended up going for the non-GADT approach together with a generic version of rank2classes in #209, see: https://github.com/advancedtelematic/quickcheck-state-machine/blob/master/src/Test/StateMachine/Types/Rank2.hs |
As per discussion in #161.
The text was updated successfully, but these errors were encountered: