aws package fails to build with master

changed milestone to %8.2.1

changed weight to 10

assigned to @bgamari

How do I reproduce this? I get

cabal install --with-ghc=/home/simonpj/5builds/HEAD-4/inplace/bin/ghc-stage2
Resolving dependencies...
cabal: Could not resolve dependencies:
trying: base-4.10.0.0/installed-4.1... (dependency of aws-0.15)
next goal: vector (dependency of aws-0.15)
rejecting: vector-0.12.0.0 (conflict: base==4.10.0.0/installed-4.1..., vector
=> base>=4.5 && <4.10)
rejecting: vector-0.11.0.0 (conflict: base==4.10.0.0/installed-4.1..., vector
=> base>=4.3 && <4.10)
trying: vector-0.10.12.3
next goal: primitive (dependency of vector-0.10.12.3)
rejecting: primitive-0.6.2.0 (conflict: vector => primitive>=0.5.0.1 &&
<0.6.2)
rejecting: primitive-0.6.1.0 (conflict: base==4.10.0.0/installed-4.1...,
primitive => base>=4.3 && <4.10)
rejecting: primitive-0.5.4.0 (conflict: base==4.10.0.0/installed-4.1...,
primitive => base>=4.3 && <4.9)
rejecting: primitive-0.5.3.0, primitive-0.5.2.1 (conflict:
base==4.10.0.0/installed-4.1..., primitive => base>=4.3 && <4.8)
rejecting: primitive-0.5.1.0 (conflict: base==4.10.0.0/installed-4.1...,
primitive => base>=4 && <4.8)
rejecting: primitive-0.5.0.1 (conflict: base==4.10.0.0/installed-4.1...,
primitive => base>=4 && <4.7)
rejecting: primitive-0.5, primitive-0.4.1, primitive-0.4.0.1, primitive-0.4,
primitive-0.3.1, primitive-0.3, primitive-0.2.1, primitive-0.2, primitive-0.1
(conflict: vector => primitive>=0.5.0.1 && <0.6.2)
rejecting: primitive-0.6 (conflict: base==4.10.0.0/installed-4.1..., primitive
=> base>=4.3 && <4.9)
Dependency tree exhaustively searched.
simonpj@cam-05-unx:~/tmp/aws-0.15$

Just plain cabal install aws --allow-newer --with-ghc=... should do it.

N.B. `--allow-newer is quite handy when testing pre-release GHC's against Cabal packages; it tells Cabal to ignore upper bounds on the packages dependencies.

I can't reproduce this with HEAD

Interesting; I have been able to reproduce this pretty reliably. I'll try to have a look.

Indeed, I can't reproduce this either with a regular GHC HEAD. I wonder if having a profiled GHC HEAD makes a difference, as I noticed in your log that you were compiling a file with the suffix .p_o.

Yep, a profiled GHC HEAD is crucial here. I couldn't reproduce this with a profiled 8.0.2, but with a profiled GHC HEAD, I get:

$ cabal build
Building aws-0.16...
Preprocessing library aws-0.16...
[69 of 78] Compiling Aws.DynamoDb.Core ( Aws/DynamoDb/Core.hs, dist/build/Aws/DynamoDb/Core.p_o )

<no location info>: error:
    ghc-stage2: panic! (the 'impossible' happened)
  (GHC version 8.1.20170213 for x86_64-unknown-linux):
        completeCall
  $wloop_length_s1bOu
  Stop[BoringCtxt] Int# -> Int# -> Int
  Call stack:
      CallStack (from HasCallStack):
        prettyCurrentCallStack, called at compiler/utils/Outputable.hs:1197:58 in ghc:Outputable
        callStackDoc, called at compiler/utils/Outputable.hs:1201:37 in ghc:Outputable
        pprPanic, called at compiler/simplCore/Simplify.hs:1598:9 in ghc:Simplify
CallStack (from -prof):
  SimplCore.SimplTopBinds (compiler/simplCore/SimplCore.hs:729:29-64)
  SimplCore.Simplify (compiler/simplCore/SimplCore.hs:426:40-55)
  HscMain.Core2Core (compiler/main/HscMain.hs:1170:7-42)
  HscMain.hscSimplify' (compiler/main/HscMain.hs:(1167,1)-(1170,42))
  HscMain.finish (compiler/main/HscMain.hs:(728,1)-(740,20))
  HscMain.hscIncrementalCompile (compiler/main/HscMain.hs:(639,1)-(694,32))
  GhcMake.upsweep_mod.compile_it (compiler/main/GhcMake.hs:(1380,13)-(1382,66))
  GhcMake.upsweep_mod (compiler/main/GhcMake.hs:(1328,1)-(1482,49))
  GhcMake.parUpsweep_one.withSem (compiler/main/GhcMake.hs:1104:13-66)
  GhcMake.parUpsweep_one (compiler/main/GhcMake.hs:(1009,1)-(1164,65))
  GhcMake.parUpsweep.\.spawnWorkers.\.\ (compiler/main/GhcMake.hs:(867,43)-(918,49))
  GhcMake.parUpsweep.\.spawnWorkers.\ (compiler/main/GhcMake.hs:(867,13)-(918,49))
  GhcMake.parUpsweep.\.spawnWorkers (compiler/main/GhcMake.hs:(866,11)-(918,49))
  GhcMake.parUpsweep.\.finallySyncSession (compiler/main/GhcMake.hs:(832,9)-(834,30))
  GhcMake.parUpsweep.\ (compiler/main/GhcMake.hs:(829,62)-(949,44))
  GhcMake.parUpsweep (compiler/main/GhcMake.hs:(796,1)-(979,36))
  GhcMake.load'.upsweep_fn (compiler/main/GhcMake.hs:(351,9)-(352,41))
  GhcMake.load'.checkHowMuch (compiler/main/GhcMake.hs:(233,9)-(235,27))
  GhcMake.load' (compiler/main/GhcMake.hs:(210,1)-(439,38))
  GhcMake.load (compiler/main/GhcMake.hs:(202,1)-(204,44))
  GHC.withCleanupSession (compiler/main/GHC.hs:(450,1)-(459,37))
  GHC.runGhc (compiler/main/GHC.hs:(425,1)-(430,26))
  GHC.defaultErrorHandler (compiler/main/GHC.hs:(365,1)-(397,7))

Please report this as a GHC bug:  http://www.haskell.org/ghc/reportabug

Ahh, yes, that would make sense. Good catch, RyanGlScott!

I've managed the problem to isolate this smaller example:

module Bug where

-------------------------------------------------------------------------------
import qualified Data.ByteString.Base64       as Base64
import qualified Data.ByteString.Char8        as B
import qualified Data.Set                     as S
import qualified Data.Text                    as T
import qualified Data.Text.Encoding           as T

data DValue
    = DBinary B.ByteString
    | DBool Bool
    | DBoolSet (S.Set Bool)

-------------------------------------------------------------------------------
class DynSize a where
    dynSize :: a -> Int

instance DynSize DValue where
    dynSize (DBool _) = 8
    dynSize (DBoolSet s) = sum $ map (dynSize . DBool) $ S.toList s
    dynSize (DBinary bs) = T.length . T.decodeUtf8 $ Base64.encode bs

You'll need to run cabal install base64-bytestring text --enable-library-profiling beforehand. Then you can reproduce it like so:

$ ~/Software/ghc3/inplace/bin/ghc-stage2 -O2 -prof -fforce-recomp Bug.hs
[1 of 1] Compiling Bug              ( Bug.hs, Bug.o )
ghc-stage2: panic! (the 'impossible' happened)
  (GHC version 8.1.20170213 for x86_64-unknown-linux):
        completeCall
  $wloop_length_s87d
  Stop[BoringCtxt] Int# -> Int# -> Int
  Call stack:
      CallStack (from HasCallStack):
        prettyCurrentCallStack, called at compiler/utils/Outputable.hs:1197:58 in ghc:Outputable
        callStackDoc, called at compiler/utils/Outputable.hs:1201:37 in ghc:Outputable
        pprPanic, called at compiler/simplCore/Simplify.hs:1598:9 in ghc:Simplify
CallStack (from -prof):
  SimplCore.SimplTopBinds (compiler/simplCore/SimplCore.hs:729:29-64)
  SimplCore.Simplify (compiler/simplCore/SimplCore.hs:426:40-55)
  HscMain.Core2Core (compiler/main/HscMain.hs:1170:7-42)
  HscMain.hscSimplify' (compiler/main/HscMain.hs:(1167,1)-(1170,42))
  HscMain.finish (compiler/main/HscMain.hs:(728,1)-(740,20))
  HscMain.hscIncrementalCompile (compiler/main/HscMain.hs:(639,1)-(694,32))
  GhcMake.upsweep_mod.compile_it (compiler/main/GhcMake.hs:(1380,13)-(1382,66))
  GhcMake.upsweep_mod (compiler/main/GhcMake.hs:(1328,1)-(1482,49))
  GhcMake.upsweep.upsweep' (compiler/main/GhcMake.hs:(1197,3)-(1296,91))
  GhcMake.upsweep (compiler/main/GhcMake.hs:(1189,1)-(1296,91))
  GhcMake.load'.upsweep_fn (compiler/main/GhcMake.hs:(351,9)-(352,41))
  GhcMake.load'.checkHowMuch (compiler/main/GhcMake.hs:(233,9)-(235,27))
  GhcMake.load' (compiler/main/GhcMake.hs:(210,1)-(439,38))
  GhcMake.load (compiler/main/GhcMake.hs:(202,1)-(204,44))
  GHC.withCleanupSession (compiler/main/GHC.hs:(450,1)-(459,37))
  GHC.runGhc (compiler/main/GHC.hs:(425,1)-(430,26))
  GHC.defaultErrorHandler (compiler/main/GHC.hs:(365,1)-(397,7))

Please report this as a GHC bug:  http://www.haskell.org/ghc/reportabug

I've confirmed that this bug first appears after commit 8d5cf8bf (Join points).

Great test case! I'm on it

assigned to @simonpj and unassigned @bgamari

mentioned in commit 2d5be63d

closed

This was fixed in 5cef996c. About to push a commit adding the testcase from ticket:13255#comment:131994.

Trac metadata

Trac field	Value
Resolution	Unresolved → ResolvedFixed
Test case	- → simplCore/should_compile/T13255

Unfortunately I hadn't previously noticed that the repro from ticket:13255#comment:131994 involved Text and bytestring-base64, neither of which are anywhere near being boot packages. I'm afraid it looks like adding a testcase for this one isn't currently possible.

Trac metadata

Trac field	Value
Test case	simplCore/should_compile/T13255 → -

Are you sure that 5cef996c is merged to HEAD? I don't see it.

reopened

Trac metadata

Trac field	Value
Resolution	ResolvedFixed → Unresolved

mentioned in commit 6e328847

closed

Indeed the commit was lost during a rebase. There we are.

Trac metadata

Trac field	Value
Resolution	Unresolved → ResolvedFixed

added compiler crash label

added Phighest label

Trac field	Value
Version	8.0.1
Type	Bug
TypeOfFailure	OtherFailure
Priority	highest
Resolution	Unresolved
Component	Compiler
Test case
Differential revisions
BlockedBy
Related
Blocking
CC
Operating system
Architecture

aws package fails to build with master

Child items ...

Activity