Denis Buzdalov
Denis Buzdalov
Wow! Moreover, if you rename `k`, say, to `j` in the implementation, all fails as expected, as well as renaming your `S` to something.
> Because this won't catch with-apps on the RHS for instance! I don't understand the case. With-apps on the RHS are coded with `IWithApp` and are showed correctly. Or, do...
Maybe, then, `--clean` action should have a variant of cleaning files of all TTC versions?
By the way, would old `depends` directories made as symlinks to, say, `../../build/ttc`, stop working with this PR?
> The test suite has examples with `depends`. Are these not satisfactory? No, they are not since no such test contains a single TTC file. Thus, we are not testing...
> Can Haskell do this? If I understand the context correctly, i.e. when we are taking about concrete stacks, Haskell does optimize them away through aggressive inlining, as it was...
I'm not sure this is relevant, but there was another issue related to `auto`-implicits and different behaviour depending on whether this was happened in the same or different modules: #814
Mutual records also have projections cross-available, when your approach, as far as I understand, forward-declares only data. Is it possible to forward-declare projections too?
> Do you have an example of the behavior you want? I believe happens with this patch should be exactly the same behavior as mutual blocks. (The code for desugaring...
> I think the same style should be used throughout the code base. If we are talking about whole `pack`, it's not a problem to have the same style throughout,...