crackNum fails to build with newer sbv

[mihaimaruseac]

Preprocessing executable 'crackNum' for crackNum-3.14..                                                                                                                                                      
       Building executable 'crackNum' for crackNum-3.14..
       [3 of 3] Compiling Main [Data.SBV package changed]

       /home/curators/work/unpack-dir/unpacked/crackNum-3.14-407180036aaabd70fe2c8cf43e2934f6b93e407010eac998dccb0b3fc85193be/src/CrackNum/Main.hs:562:40: error: [GHC-39999]
           • Ambiguous type variable ‘a1’ arising from a use of ‘.==’
             prevents the constraint ‘(SymVal a1)’ from being solved.
             Probable fix: use a type annotation to specify what ‘a1’ should be.
             Potentially matching instances:
               instance (SymVal a, SymVal b) => SymVal (Either a b)
                 -- Defined in ‘sbv-11.0:Data.SBV.Core.Model’
               instance SymVal Int16 -- Defined in ‘sbv-11.0:Data.SBV.Core.Model’
               ...plus 30 others
               (use -fprint-potential-instances to see them all)
           • In the second argument of ‘($)’, namely ‘SBV x .== v’
             In a stmt of a 'do' block: pure $ SBV x .== v
             In the expression:
               do let k = KBounded sgn n
                      v = SBV $ SVal k $ Left $ mkConstCV k iv
                  x <- (if sgn then sIntN else sWordN) n "ENCODED"
                  pure $ SBV x .== v
           |
       562 |                           pure $ SBV x .== v
           |                                        ^^^

Placing an upper bound on sbv for now for a few weeks, per https://github.com/commercialhaskell/stackage/issues/7568#issuecomment-2461288085

CC @LeventErkok

[LeventErkok]

@mihaimaruseac Which version of GHC are you using for this?

Because with GHC9.10.1, I can compile it just fine against the latest SBV (v11) without any issues.

[mihaimaruseac]

This is stackage nightly, which uses ghc 9.8.3: https://www.stackage.org/nightly-2024-11-08

[LeventErkok]

I don't have a ghc 9.8.3 setup to try this out; so I guess this one won't make it to stackage. Oh well.

[mihaimaruseac]

LTS is currently on 9.6.6. The next LTS will start from nightly at the time we start it (probably on 9.8.4?).

Will likely have to do similar bounds for LTS

[LeventErkok]

Guess all I'm saying is I won't be uploading a new crackNum to make this problem go away.. How Stackage deals with that, I'm not sure.

[phadej]

@LeventErkok you can ghcup install ghc 9.8.2 (9.8.3 is not in ghcup yet) and cabal build -w ghc-9.8.2.

That said, I can also reproduce this issue on GHC-9.10.1:

crackNum-3.14 % cabal build -w ghc-9.10.1
Resolving dependencies...
Build profile: -w ghc-9.10.1 -O1
In order, the following will be built (use -v for more details):
 - crackNum-3.14 (exe:crackNum) (first run)
Configuring executable 'crackNum' for crackNum-3.14..
Preprocessing executable 'crackNum' for crackNum-3.14..
Building executable 'crackNum' for crackNum-3.14..
[1 of 3] Compiling CrackNum.TestSuite ( src/CrackNum/TestSuite.hs, /codetmp/crackNum-3.14/dist-newstyle/build/x86_64-linux/ghc-9.10.1/crackNum-3.14/x/crackNum/build/crackNum/crackNum-tmp/CrackNum/TestSuite.o )
[2 of 3] Compiling Paths_crackNum   ( /codetmp/crackNum-3.14/dist-newstyle/build/x86_64-linux/ghc-9.10.1/crackNum-3.14/x/crackNum/build/crackNum/autogen/Paths_crackNum.hs, /codetmp/crackNum-3.14/dist-newstyle/build/x86_64-linux/ghc-9.10.1/crackNum-3.14/x/crackNum/build/crackNum/crackNum-tmp/Paths_crackNum.o )
[3 of 3] Compiling Main             ( src/CrackNum/Main.hs, /codetmp/crackNum-3.14/dist-newstyle/build/x86_64-linux/ghc-9.10.1/crackNum-3.14/x/crackNum/build/crackNum/crackNum-tmp/Main.o )
src/CrackNum/Main.hs:562:40: error: [GHC-39999]
    • Ambiguous type variable ‘a1’ arising from a use of ‘.==’
      prevents the constraint ‘(SymVal a1)’ from being solved.
      Probable fix: use a type annotation to specify what ‘a1’ should be.
      Potentially matching instances:
        instance (SymVal a, SymVal b) => SymVal (Either a b)
          -- Defined in ‘sbv-11.0:Data.SBV.Core.Model’
        instance SymVal Int16 -- Defined in ‘sbv-11.0:Data.SBV.Core.Model’
        ...plus 30 others
        (use -fprint-potential-instances to see them all)
    • In the second argument of ‘($)’, namely ‘SBV x .== v’
      In a stmt of a 'do' block: pure $ SBV x .== v
      In the expression:
        do let k = KBounded sgn n
               v = SBV $ SVal k $ Left $ mkConstCV k iv
           x <- (if sgn then sIntN else sWordN) n "ENCODED"
           pure $ SBV x .== v
    |
562 |                           pure $ SBV x .== v
    |                                        ^^^

built against sbv-0.11, libBF-0.6.8

[phadej]

FWIW, I also noticed that some sbv version have incorrect lower bound on base, e.g. 10.12 with ghc-9.2.8:

Data/SBV/Core/Model.hs:64:1: error: [-Wdodgy-imports, -Werror=dodgy-imports]
    Module ‘GHC.TypeLits’ does not export ‘SChar’
   |
64 | import GHC.TypeLits hiding (SChar)
   | ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

I'll make revisions later today.

[LeventErkok]

I can’t replicate the failure with GHC 9.10.1, SBV 11.0 and CrackNum 3.14. Compiles just fine on my Mac.

I’m at a loss to how to debug this.

[LeventErkok]

Oh, never mind.. I blew-up my cabal cache and did a rebuild and can now see the failure. My installation must've gotten to a bizarre state.

I'll push a fix soon. Sorry about the noise.

[LeventErkok]

CrackNum 3.15 is now on Hackage: https://hackage.haskell.org/package/crackNum-3.15

It should compile cleanly with SBV 11.0

[mihaimaruseac]

Thank you