haskell-distributed / rank1dynamic

Like Data.Dynamic and Data.Typeable but supporting polymorphic values

rank1dynamic

See http://haskell-distributed.github.com for documentation, user guides, tutorials and assistance.

Getting Help / Raising Issues

Please visit the bug tracker to submit issues. You can contact the [email protected] mailing list for help and comments.

License

This package is made available under a 3-clause BSD-style license.

rank1dynamic's Issues

Release version 0.3.1.0

Checklist
[ ] Update cabal file
[ ] Signed tag commit
[ ] Upload to hackage

Test failure with ghc-9.2.4 (FUN instead of `->`)

It seems like the tests are failing with ghc-9.2.4 with this error:

  CANNOT use a term of type 'forall a. a -> a' as 'forall a. a': [Failed]
expected: Left "Cannot unify Skolem and ->"
 but got: Left "Cannot unify Skolem and FUN"

Should we add a new funKindStr value in https://github.com/haskell-distributed/rank1dynamic/blob/master/tests/test.hs#L10-L15 ?

build failed on Nix server

i was trying to use distributed-process-0.7.4 with nix but it seems that build is failing https://hydra.nixos.org/build/86431664 because of rank1dynamic build is failing.

please take a look on the logs https://hydra.nixos.org/build/86437016/log

let me know if there is an easy way to fix or workaround this.

Bump HUnit dependency

Builds with HUnit-1.5.0.0 and otherwise breaks the NixOS build (https://hydra.nixos.org/build/49484130)

GHC 8.2.1 Typeable changes cause compilation failure

Data.Typeable has changed significantly with GHC 8.2.1.

src/Data/Rank1Typeable.hs:107:5: error:
    Module ‘Data.Typeable’ does not export ‘mkTyCon3’
    |
107 |   , mkTyCon3
    |     ^^^^^^^^

src/Data/Rank1Typeable.hs:115:1: error:
    Could not find module ‘Data.Typeable.Internal’
    it is a hidden module in the package ‘base-4.10.0.0’
    Use -v to see a list of the files searched for.
    |
115 | import qualified Data.Typeable.Internal as T
    | ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

src/Data/Rank1Typeable.hs:121:5: error:
    Module ‘Data.Typeable’ does not export ‘mkTyConApp’
    |
121 |   , mkTyConApp
    |     ^^^^^^^^^^

It looks like there may be GHC 8.2.1-specific macro blocks needed.

Bad base version bounds on some versions

See https://matrix.hackage.haskell.org/package/rank1dynamic

compatibility with ghc-8

src/Data/Rank1Typeable.hs:113:32: error:
    Module ‘Data.Typeable.Internal’ does not export ‘tcList’

src/Data/Rank1Typeable.hs:113:40: error:
    Module ‘Data.Typeable.Internal’ does not export ‘tcFun’

Btw, the "official" bug tracker did not work in my firefox, effectively threw away my filled out issue form because i was not logged in (which is a wild guess; it did not allow me to submit the issue with some "no reporter" error, but there is no "reporter" field on the form) and then refused to let me register with one of my usual mail addresses.

Thanks for providing this package.

New release?

Could we please have a new release? In particular I'd like the fix for #24 released so that I can use a package which depends on distributed-process on nix.

Compile error on GHC 8.0 (base 4.9.0)

Before removing version contraints:

[nix-shell:~/workspace/rank1dynamic]$ runhaskell Setup.hs  configure
Configuring rank1dynamic-0.3.1.0...
Setup.hs: At least the following dependencies are missing:
binary >=0.5 && <0.8, ghc-prim >=0.2 && <0.5

After removing version constraints:

Building rank1dynamic-0.3.1.0...
Preprocessing library rank1dynamic-0.3.1.0...
[1 of 2] Compiling Data.Rank1Typeable ( src/Data/Rank1Typeable.hs, dist/build/Data/Rank1Typeable.o )

src/Data/Rank1Typeable.hs:105:32: error:
    Module ‘Data.Typeable.Internal’ does not export ‘listTc’

src/Data/Rank1Typeable.hs:105:47: error:
    Module ‘Data.Typeable.Internal’ does not export ‘TyCon(TyCon)’
<3>builder for ‘/nix/store/szdn6wp7al4pxcay3lparc41wdsfnhlq-rank1dynamic-0.3.1.0.drv’ failed with exit code 1

My environment:

[nix-shell:~/workspace/rank1dynamic]$ ghc-pkg list
/nix/store/xirfbpcf0yb6b511r93qzm4y83gf9xqw-ghc-8.0.0.20160111/lib/ghc-8.0.0.20160111/package.conf.d
    Cabal-1.23.0.0
    HUnit-1.3.1.0
    ansi-terminal-0.6.2.3
    ansi-wl-pprint-0.6.7.3
    array-0.5.1.0
    base-4.9.0.0
    binary-0.8.0.0
    bytestring-0.10.7.0
    containers-0.5.7.1
    deepseq-1.4.2.0
    directory-1.2.5.0
    extensible-exceptions-0.1.1.4
    filepath-1.4.1.0
    ghc-8.0.0.20160111
    ghc-boot-8.0.0.20160111
    ghc-prim-0.5.0.0
    ghci-8.0.0.20160111
    haskeline-0.7.2.2
    hoopl-3.10.2.1
    hostname-1.0
    hpc-0.6.0.3
    integer-gmp-1.0.0.1
    mtl-2.2.1
    old-locale-1.0.0.7
    pretty-1.1.3.2
    process-1.4.1.0
    random-1.1
    regex-base-0.93.2
    regex-posix-0.95.2
    rts-1.0
    template-haskell-2.11.0.0
    terminfo-0.4.0.2
    test-framework-0.8.1.1
    test-framework-hunit-0.3.0.2
    text-1.2.2.0
    time-1.6
    transformers-0.5.0.0
    unix-2.7.1.1
    xhtml-3000.2.1
    xml-1.3.14

[nix-shell:~/workspace/rank1dynamic]$ ghc --version
The Glorious Glasgow Haskell Compilation System, version 8.0.0.20160111

Higher-kinded type variables?

I find myself wanting to be able to Typeable quantify over higher-kinded variables, like * -> * or Constraint. This problem comes in two parts: one of them I've thought a lot about and I'm pretty sure I've cracked, and one of them I haven't been able to swing at successfully.

{-# LANGUAGE PolyKinds, DataKinds, EmptyDataDecls, MultiParamTypeClasses, UndecidableInstances, UndecidableSuperclasses #-}
import GHC.TypeLits (Nat)
import qualified GHC.Exts as E (Any)

There's two different approaches, which depend on what version of GHC you have. The first is the simpler, if you have a newer version.

data family Any :: Nat -> k -- Not exported
type family ANY :: Nat -> k where -- Exported
  ANY = Any

Unlike the type family Any from GHC.Exts, the data family here is Typeable.

The other works if you have an older version which can't declare data families ending in anything other than *.

data AnyT0 (n :: Nat) -- None of these datatypes are exported.
data AnyT1 (n :: Nat) (a :: k)
data AnyT2 (n :: Nat) (a :: ka) (b :: kb)
-- And so on.
class E.Any => AnyC0 (n :: Nat) -- And neither are these.
class E.Any => AnyC1 (n :: Nat) (a :: k)
class E.Any => AnyC2 (n :: Nat) (a :: ka) (b :: kb)
type family ANY :: Nat -> k where -- This is, though.
  ANY = AnyT0
  ANY = AnyT1
  ANY = AnyT2
  ANY = AnyC0
  ANY = AnyC1
  ANY = AnyC2
  -- And so on.

This is also more convenient because you use GHC-provided Nats rather than bespoke Peano nats.

The problem I'm having, and the reason this is an issue rather than a pull request, is I don't know how to do inference and unification on these types. I suspect you'd need something similar to GHC's code for it.