May 2017 - ghc-tickets

[GHC] #13354: Package database locking patch broke ghc-pkg with non-existent database
by GHC 23 Jul '17

23 Jul '17

#13354: Package database locking patch broke ghc-pkg with non-existent database -------------------------------------+------------------------------------- Reporter: bgamari | Owner: (none) Type: bug | Status: new Priority: highest | Milestone: 8.2.1 Component: Package | Version: 8.0.1 system | Keywords: | Operating System: Unknown/Multiple Architecture: | Type of failure: None/Unknown Unknown/Multiple | Test Case: | Blocked By: Blocking: | Related Tickets: Differential Rev(s): | Wiki Page: -------------------------------------+------------------------------------- It seems that Phab:D3090 (#13194) broke `ghc-pkg`'s behavior in the face of non-existent package databases. Namely, `cabal install` fails during registration with, {{{ Creating package registration file: /tmp/pkgConf- compact-0.1.0.1-13684/pkgConf ("/opt/exp/ghc/roots/master/bin/ghc-pkg",["update","-","--global","-- user","-v2"]) /opt/exp/ghc/roots/master/bin/ghc-pkg returned ExitFailure 1 with error message: ghc-pkg: Couldn't open database /home/ben/.ghc/x86_64-linux-8.1.20170227/package.conf.d for modification: /home/ben/.ghc/x86_64-linux-8.1.20170227/package.conf.d.lock: openBinaryFile: does not exist (No such file or directory) }}} -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/13354> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 7

[GHC] #11375: Type aliases twice as slow to compile as closed type families.
by GHC 22 Jul '17

22 Jul '17

#11375: Type aliases twice as slow to compile as closed type families. -------------------------------------+------------------------------------- Reporter: danilo2 | Owner: Type: bug | Status: new Priority: high | Milestone: Component: Compiler | Version: 8.1 (Type checker) | Keywords: | Operating System: MacOS X Architecture: aarch64 | Type of failure: Compile-time | performance bug Test Case: | Blocked By: Blocking: | Related Tickets: Differential Rev(s): | Wiki Page: -------------------------------------+------------------------------------- Hello guys! This bug is awkard at last to me. So I notiiced today, that using type aliases in my code takes twice as long to copile than using closed type families with just single member. So to be most precise - when I replace the following line: {{{ type XQ1 m a = Targets (Match m a) }}} with {{{ type family XQ1 m a where XQ1 m a = Targets (Match m a) }}} the compilation time drops from 15s to 7s (this type alias is heavily used across the application). I think it could be somehow related to the famous #8095 bug, but I can of course be wrong here. Unfortunetally the codebase is pretty big and its not so easy to make a minimal example, but I'll try to work on it during the upcoming weekend (please notify me if that would not be necessary). And some statistics from `-dshow-passes`: When using type alias: {{{ *** Parser: *** Renamer/typechecker: *** Desugar: Result size of Desugar (after optimization) = {terms: 10,303, types: 66,949, coercions: 16,383,834} *** Simplifier: Result size of Simplifier iteration=1 = {terms: 10,722, types: 75,391, coercions: 16,507,338} Result size of Simplifier iteration=2 = {terms: 10,550, types: 74,437, coercions: 16,504,770} Result size of Simplifier iteration=3 = {terms: 10,543, types: 74,347, coercions: 16,504,734} Result size of Simplifier = {terms: 10,543, types: 74,347, coercions: 16,504,630} *** Tidy Core: Result size of Tidy Core = {terms: 10,846, types: 75,522, coercions: 16,504,729} *** CorePrep: Result size of CorePrep = {terms: 14,823, types: 96,430, coercions: 16,504,729} *** ByteCodeGen: Upsweep completely successful. *** Deleting temp files: Warning: deleting non-existent /var/folders/y9/km24zv3d4yl711mwv85vwljc0000gp/T/ghc64965_0/ghc_95.c Warning: deleting non-existent /var/folders/y9/km24zv3d4yl711mwv85vwljc0000gp/T/ghc64965_0/ghc_94.o Ok, modules loaded: Prologue, Type.Bool, Type.Container, Type.List, Type.Set, Data.Cata, Type.Promotion, Data.Bits.Mask, Control.Monad.State.Dependent, Control.Monad.Poly, Control.Applicative.Poly, Type.Regex, Type.Either, Data.Reprx, Data.Impossible, Type.Cache.TH, Main, Control.Lens.Utils, Control.Lens.Wrapped.Utils, Data.Text.CodeBuilder, Data.Text.CodeBuilder.Tok, Data.Text.CodeBuilder.Builder, Data.Text.CodeBuilder.Doc, Data.Binary.Instances.Missing, Data.Default.Instances.Missing, Data.String.Class, Data.Text.Class, Data.Convert, Data.Layer, Data.Container.Class, Data.Container.List, Data.Functor.Utils, Type.Operators, Type.Show, Type.Wrapped, Data.Container.Opts, Data.Container.Poly, Data.Convert.Base, Data.Convert.Instances, Data.Convert.Instances.Map, Data.Convert.Instances.Num, Data.Convert.Instances.Text, Data.Convert.Instances.TH, Data.Convert.Bound, Data.Bits.Base. *** Parser: *** Desugar: *** Simplify: *** CorePrep: *** ByteCodeGen: ... }}} and with type families: {{{ *** Parser: *** Renamer/typechecker: *** Desugar: Result size of Desugar (after optimization) = {terms: 310, types: 867, coercions: 78} *** Simplifier: Result size of Simplifier iteration=1 = {terms: 301, types: 878, coercions: 296} Result size of Simplifier = {terms: 301, types: 878, coercions: 293} *** Tidy Core: Result size of Tidy Core = {terms: 337, types: 1,002, coercions: 293} *** CorePrep: Result size of CorePrep = {terms: 475, types: 1,305, coercions: 293} *** ByteCodeGen: *** Deleting temp files: Warning: deleting non-existent /var/folders/y9/km24zv3d4yl711mwv85vwljc0000gp/T/ghc64863_0/ghc_93.c Warning: deleting non-existent /var/folders/y9/km24zv3d4yl711mwv85vwljc0000gp/T/ghc64863_0/ghc_92.o compile: input file /var/folders/y9/km24zv3d4yl711mwv85vwljc0000gp/T/ghc64863_0/ghc_1.hscpp *** Checking old interface for Main: [45 of 45] Compiling Main ( /Users/wdanilo/dev/hs/public/variants/Main.hs, interpreted ) *** Parser: *** Renamer/typechecker: *** Desugar: Result size of Desugar (after optimization) = {terms: 10,303, types: 67,609, coercions: 162,249} *** Simplifier: Result size of Simplifier iteration=1 = {terms: 10,722, types: 75,721, coercions: 323,027} Result size of Simplifier iteration=2 = {terms: 10,550, types: 74,767, coercions: 320,473} Result size of Simplifier iteration=3 = {terms: 10,543, types: 74,677, coercions: 320,437} Result size of Simplifier = {terms: 10,543, types: 74,677, coercions: 320,333} *** Tidy Core: Result size of Tidy Core = {terms: 10,846, types: 75,852, coercions: 320,432} *** CorePrep: Result size of CorePrep = {terms: 14,823, types: 96,760, coercions: 320,432} *** ByteCodeGen: Upsweep completely successful. *** Deleting temp files: Warning: deleting non-existent /var/folders/y9/km24zv3d4yl711mwv85vwljc0000gp/T/ghc64863_0/ghc_95.c Warning: deleting non-existent /var/folders/y9/km24zv3d4yl711mwv85vwljc0000gp/T/ghc64863_0/ghc_94.o Ok, modules loaded: Prologue, Type.Bool, Type.Container, Type.List, Type.Set, Data.Cata, Type.Promotion, Data.Bits.Mask, Control.Monad.State.Dependent, Control.Monad.Poly, Control.Applicative.Poly, Type.Regex, Type.Either, Data.Reprx, Data.Impossible, Type.Cache.TH, Main, Control.Lens.Utils, Control.Lens.Wrapped.Utils, Data.Text.CodeBuilder, Data.Text.CodeBuilder.Tok, Data.Text.CodeBuilder.Builder, Data.Text.CodeBuilder.Doc, Data.Binary.Instances.Missing, Data.Default.Instances.Missing, Data.String.Class, Data.Text.Class, Data.Convert, Data.Layer, Data.Container.Class, Data.Container.List, Data.Functor.Utils, Type.Operators, Type.Show, Type.Wrapped, Data.Container.Opts, Data.Container.Poly, Data.Convert.Base, Data.Convert.Instances, Data.Convert.Instances.Map, Data.Convert.Instances.Num, Data.Convert.Instances.Text, Data.Convert.Instances.TH, Data.Convert.Bound, Data.Bits.Base. *** Parser: *** Desugar: *** Simplify: *** CorePrep: *** ByteCodeGen: ... }}} As we can see the type aliases produce much more coertions: 16,383,834 vs 78 using type families. This is really interesting thing and it seems related to the linked bug above. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/11375> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 8

[GHC] #13680: Can't use TypeApplications with empty list expression
by GHC 19 Jul '17

19 Jul '17

#13680: Can't use TypeApplications with empty list expression -------------------------------------+------------------------------------- Reporter: RyanGlScott | Owner: (none) Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 8.0.1 (Type checker) | Keywords: | Operating System: Unknown/Multiple TypeApplications | Architecture: | Type of failure: GHC rejects Unknown/Multiple | valid program Test Case: | Blocked By: Blocking: | Related Tickets: Differential Rev(s): | Wiki Page: -------------------------------------+------------------------------------- Why can't I do this? {{{#!hs {-# LANGUAGE TypeApplications #-} module Bug where foo :: [Int] foo = [] @Int }}} Compiling this with GHC 8.0.1, 8.0.2, 8.2.1, or HEAD gives me: {{{ $ /opt/ghc/head/bin/ghci Bug2.hs GHCi, version 8.3.20170509: http://www.haskell.org/ghc/ :? for help Loaded GHCi configuration from /home/rgscott/.ghci [1 of 1] Compiling Bug ( Bug2.hs, interpreted ) Bug2.hs:5:7: error: • Cannot apply expression of type ‘[a0]’ to a visible type argument ‘Int’ • In the expression: [] @Int In an equation for ‘foo’: foo = [] @Int | 5 | foo = [] @Int | ^^^^^^^ }}} This seems really strange, since I can use `TypeApplications` with expressions like `Nothing @Int` without issues. According to GHCi: {{{ $ /opt/ghc/head/bin/ghci GHCi, version 8.3.20170509: http://www.haskell.org/ghc/ :? for help Loaded GHCi configuration from /home/rgscott/.ghci λ> :set -XTypeApplications -fprint-explicit-foralls λ> :type +v [] [] :: forall {a}. [a] λ> :type +v Nothing Nothing :: forall a. Maybe a }}} The type variable for `[]` isn't visible! But I don't see any reason why it shouldn't be, especially since, conceptually, the data type declaration for lists is: {{{#!hs data [] a = [] | a : [a] }}} I suspect that `[]`'s tyvar binder visibility is not wired into GHC correctly. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/13680> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 15

[GHC] #13434: hs_try_putmvar003 is timing out / segfaulting
by GHC 19 Jul '17

19 Jul '17

#13434: hs_try_putmvar003 is timing out / segfaulting -------------------------------------+------------------------------------- Reporter: ezyang | Owner: (none) Type: bug | Status: new Priority: normal | Milestone: Component: Runtime | Version: 8.1 System | Keywords: | Operating System: Unknown/Multiple Architecture: | Type of failure: Runtime crash Unknown/Multiple | Test Case: | Blocked By: Blocking: | Related Tickets: Differential Rev(s): | Wiki Page: -------------------------------------+------------------------------------- This is basically reproduceable every time I validate. During validation, the program times out, but when I run it by hand, it segfaults. Running in debug sanity mode, I get: {{{ hstry: internal error: MVAR_CLEAN on mutable list (GHC version 8.3.20170316 for x86_64_unknown_linux) Please report this as a GHC bug: http://www.haskell.org/ghc/reportabug }}} Simon Marlow, I'm happy to investigate more if you can't repro. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/13434> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 11

[GHC] #11095: -O0 -g slows GHC down on list literals (compared to -O0 without -g)
by GHC 19 Jul '17

19 Jul '17

#11095: -O0 -g slows GHC down on list literals (compared to -O0 without -g) -------------------------------------+------------------------------------- Reporter: slyfox | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 7.10.2 Keywords: | Operating System: Unknown/Multiple Architecture: | Type of failure: None/Unknown Unknown/Multiple | Test Case: | Blocked By: Blocking: | Related Tickets: Differential Rev(s): | Wiki Page: -------------------------------------+------------------------------------- Tried to buld syb-0.6 testsuite with '-O0 -g' flags today and noticed compiler hung for a couple of hours. Distilled example from syb-0.6 testsuite (exact example attached): {{{#!hs {-# LANGUAGE NoImplicitPrelude #-} {-# OPTIONS_GHC -O0 #-} module Bug (bug) where {-# NOINLINE f #-} f :: a -> a f v = v bug :: [()] bug = f v v :: [()] v = [(),(),(),(),(),(),(),(),(),(),(),(),(),(),(),(),(),(),(),(),(),(),(),(),(),(),(),(),(),(),(),(),(),(),(),(),(),(),(),() ,(),(),(),(),() ... --- a lot of these ] }}} {{{ $ time ghc -c -fforce-recomp -O0 Bug.hs real 0m0.333s user 0m0.288s sys 0m0.042s $ time ghc -c -fforce-recomp -O0 -g Bug.hs real 0m40.241s user 0m40.021s sys 0m0.070s }}} For some reason '''-g''' manages slow things down at a large factor. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/11095> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 20

[GHC] #8903: Add dead store elimination
by GHC 18 Jul '17

18 Jul '17

#8903: Add dead store elimination ------------------------------------+------------------------------------- Reporter: tibbe | Owner: Type: feature request | Status: new Priority: normal | Milestone: Component: Compiler | Version: 7.9 Keywords: | Operating System: Unknown/Multiple Architecture: Unknown/Multiple | Type of failure: None/Unknown Difficulty: Unknown | Test Case: Blocked By: | Blocking: Related Tickets: | ------------------------------------+------------------------------------- We could use some dead store elimination in the code generator. Here's some Cmm that has redundant stores to the same locations: {{{ // thawArray#: I64[Hp - 168] = I64[PicBaseReg + stg_MUT_ARR_PTRS_DIRTY_info@GOTPCREL]; I64[Hp - 160] = 16; I64[Hp - 152] = 17; _c2nT::I64 = Hp - 168; call MO_Memcpy(_c2nT::I64 + 24, _s2cx::P64 + 24, 128, 8); // writeArray#: P64[(_c2nT::I64 + 24) + (_s2cy::I64 << 3)] = _s2cE::P64; I64[_c2nT::I64] = I64[PicBaseReg + stg_MUT_ARR_PTRS_DIRTY_info@GOTPCREL]; I8[(_c2nT::I64 + 24) + ((I64[_c2nT::I64 + 8] << 3) + (_s2cy::I64 >> 7))] = 1 :: W8; // unsafeFreeze# I64[_c2nT::I64] = I64[PicBaseReg + stg_MUT_ARR_PTRS_FROZEN0_info@GOTPCREL]; }}} There are three stores to the same location (`I64[_c2nT::I64]`). (There's also one much less obvious double store to another location, which will probably be much harder to address: the store to `P64[(_c2nT::I64 + 24) + (_s2cy::I64 << 3)]` overwrites a word previous written by the `MO_Memcpy`. Getting to that one will be hard as the memcoy callish !MachOp only gets expanded in the backend.) -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/8903> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 4

[GHC] #9630: compile-time performance regression (probably due to Generics)
by GHC 18 Jul '17

18 Jul '17

#9630: compile-time performance regression (probably due to Generics) -------------------------------------+------------------------------------- Reporter: hvr | Owner: simonpj Type: bug | Status: new Priority: normal | Milestone: 7.10.1 Component: Compiler | Version: 7.9 Keywords: | Operating System: Architecture: Unknown/Multiple | Unknown/Multiple Difficulty: Unknown | Type of failure: Compile- Blocked By: | time performance bug Related Tickets: #9583 | Test Case: | Blocking: | Differential Revisions: -------------------------------------+------------------------------------- This is a continuation of #9583 Compiling latest Cabal from Git's `Language.Haskell.Extension` module shows a significant performance regression with respect to GHC 7.8.3: {{{ $ uname -a Linux duo 3.17.0-031700rc6-generic #201409211935 SMP Sun Sep 21 23:37:11 UTC 2014 x86_64 x86_64 x86_64 GNU/Linux $ git describe cabal-install_merged-1912-g7161e4c ######################################################################################## $ /opt/ghc/7.8.3/bin/ghc --version The Glorious Glasgow Haskell Compilation System, version 7.8.3 $ /opt/ghc/7.8.3/bin/ghc -Rghc-timing -O -c Language/Haskell/Extension.hs <<ghc: 5228751760 bytes, 543 GCs, 34135677/114940864 avg/max bytes residency (14 samples), 273M in use, 0.00 INIT (0.03 elapsed), 4.96 MUT (5.28 elapsed), 3.26 GC (3.47 elapsed) :ghc>> ######################################################################################## $ /opt/ghc/head/bin/ghc --version The Glorious Glasgow Haskell Compilation System, version 7.9.20140923 $ /opt/ghc/head/bin/ghc -Rghc-timing -O -c Language/Haskell/Extension.hs <<ghc: 125917854120 bytes, 3181 GCs, 349136895/833496072 avg/max bytes residency (30 samples), 2096M in use, 0.001 INIT (0.001 elapsed), 133.003 MUT (142.730 elapsed), 73.241 GC (78.234 elapsed) :ghc>> }}} -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9630> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 66

[GHC] #13310: Cut a new array release
by GHC 18 Jul '17

18 Jul '17

#13310: Cut a new array release -------------------------------------+------------------------------------- Reporter: bgamari | Owner: (none) Type: task | Status: new Priority: normal | Milestone: 8.2.1 Component: Compiler | Version: 8.0.1 Keywords: | Operating System: Unknown/Multiple Architecture: | Type of failure: None/Unknown Unknown/Multiple | Test Case: | Blocked By: Blocking: | Related Tickets: Differential Rev(s): | Wiki Page: -------------------------------------+------------------------------------- The `array` package needs a new release for 8.2. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/13310> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 4

[GHC] #12131: Can't solve constraints with UndecidableSuperClasses but can infer kind (+ undesired order of kinds)
by GHC 18 Jul '17

18 Jul '17

#12131: Can't solve constraints with UndecidableSuperClasses but can infer kind (+ undesired order of kinds) -------------------------------------+------------------------------------- Reporter: Iceland_jack | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 8.0.1 Keywords: | Operating System: Unknown/Multiple UndecidableSuperClasses | Architecture: | Type of failure: None/Unknown Unknown/Multiple | Test Case: | Blocked By: Blocking: | Related Tickets: #11480 #12025 Differential Rev(s): | Wiki Page: -------------------------------------+------------------------------------- The Glorious Glasgow Haskell Compilation System, version 8.0.1. Code taken from [https://gist.github.com/ekmett/b26363fc0f38777a637d gist] with {{{#!hs fmap :: p a b -> q (f a) (f b) }}} replaced by {{{#!hs fmap :: Dom f a b -> Cod f (f a) (f b) }}} If you enable `DataKinds` and `TypeInType` `:kind` will happily tell you its kind: {{{ ghci> :kind 'Main.Nat 'Main.Nat :: forall j i (p :: Cat i) (q :: Cat j) (f :: i -> j) (g :: i -> j). (FunctorOf p q f, FunctorOf p q g) => (forall (a :: i). Ob p a => q (f a) (g a)) -> Main.Nat p q f g }}} but if you try `:type Nat` it spits out a long list of unsolved constraints {{{ ghci> :type Nat <interactive>:1:1: error: solveWanteds: too many iterations (limit = 1) Unsolved: WC {wc_simple = [W] $dFunctor_ag3Y :: Main.Functor f_ag3S (CDictCan) [W] $dFunctor_ag4d :: Main.Functor g_ag3T (CDictCan) [D] _ :: Main.Functor s_ag51 (CDictCan) [D] _ :: Main.Functor s_ag5a (CDictCan) [D] _ :: Main.Functor s_ag4W (CDictCan) [D] _ :: Main.Functor s_ag55 (CDictCan) [D] _ :: Category s_ag51 (CDictCan) [D] _ :: Category s_ag5a (CDictCan) [D] _ :: Category s_ag4W (CDictCan) [D] _ :: Category s_ag55 (CDictCan) [W] hole{ag4Y} :: Dom g_ag3T ~ Dom f_ag3S (CNonCanonical) [W] hole{ag53} :: Cod g_ag3T ~ Cod f_ag3S (CNonCanonical) [D] _ :: Dom (Dom f_ag3S) ~ Op (Dom f_ag3S) (CNonCanonical) [D] _ :: Cod (Dom f_ag3S) }}} This seems like #11480. This makes undecidable superclasses a harsh master and raises two questions that I'll bundle together: 1. Why can the `:kind` be inferred but not the `:type`? It works when given extra information: {{{ ghci> :type Nat @_ @_ @(->) @(->) Nat @_ @_ @(->) @(->) :: (Cod g ~ (->), Dom g ~ (->), Cod f ~ (->), Dom f ~ (->), Main.Functor g, Main.Functor f) => (forall a. Vacuous (->) a => f a -> g a) -> Main.Nat (->) (->) f g }}} {{{ ghci> :type Nat @_ @_ @_ @_ @[] @Maybe Nat @_ @_ @_ @_ @[] @Maybe :: (forall a. Vacuous (->) a => [a] -> Maybe a) -> Main.Nat (->) (->) [] Maybe }}} {{{ ghci> :type Nat @_ @_ @_ @_ @[] @Maybe listToMaybe Nat @_ @_ @_ @_ @[] @Maybe listToMaybe :: Main.Nat (->) (->) [] Maybe }}} 2. Why is `j` positioned before `i` (most likely because of `forall a. ...` but I tried tinkering without success? Is there any trick to ordering it as you'd expect `forall i j (p :: Cat i) (q :: Cat j)` or is it not possible per #12025? -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/12131> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 7

[GHC] #9805: Use TrieMaps to speed up type class instance lookup
by GHC 17 Jul '17

17 Jul '17

#9805: Use TrieMaps to speed up type class instance lookup -------------------------------------+------------------------------------- Reporter: ezyang | Owner: ezyang Type: task | Status: new Priority: normal | Milestone: Component: Compiler (Type | Version: 7.9 checker) | Operating System: Keywords: | Unknown/Multiple Architecture: Unknown/Multiple | Type of failure: Difficulty: Unknown | None/Unknown Blocked By: | Test Case: Related Tickets: | Blocking: | Differential Revisions: -------------------------------------+------------------------------------- Currently, type class instance resolution is performed by doing a map lookup by type class, and then linearly matching against every instance. This is not great, and for a while, simonpj has been keen on using the TrieMap data structure in GHC, which has been previously used to implement a map from CoreExprs to various things (e.g. for CSE). What makes this a little tricky is that instance lookup isn't intended to be an exact match: we should unify so-called template variables and provide a substitution; furthermore, there may be multiple matches. After some prototyping, I think we should be able to make this work. The primary refactoring we have to do is *maintain the key associated with an entry in a TrieMap*. Unlike the current uses of TrieMaps, where it's acceptable to lose the underlying key associated with an entry in the TrieMap, we need to know the key when doing unification, because it may be reported in the substitution. There are a few knock-on effects of this: * We should add a method `foldWithKeyTM :: (Key m -> a -> b -> b) -> m a -> b -> b` to the `TrieMap` type class. * We need a variant of `UniqFM` which tracks the original key that was used. I propose we name it `UniqKM` (unique key map). A number of implementations of `TrieMap` need to be adjusted to use this instead of `UniqFM`. * Why can't we just represent keyed TrieMaps as `TypeMap (Type, a)`? It might be possible. An insurmountable difficulty would be if we need to know the partial key PRIOR to having finished traversing the TrieMap; however, for the parts of the unification algorithm I've implemented, this does not seem to be the case. The primary actual difficulty is that `Type` uses a named representation, whereas `TypeMap` keys are on-the-fly deBruijn numbered. There would at least be some annoying fiddliness. * Reversing the deBruijn numbered key into a `Type` is a bit goofy. Essentially, you need the reverse of the current `CmEnv`: a mapping from de Bruijn levels to the `Var` you've decided to allocate. (I've called this `CfEnv` in my prototype) * When we represent a TrieMap binder, we have to put the binder map on the OUTSIDE, as opposed to the inside as it is currently. Otherwise, we can't update the `CfEnv` with the new mapping before making the recursive call to fold-with-key. I'll attach the standalone Haskell file I used to prototype this, wherein I copy-pasted a lot of Haskell from GHC's source and implemented fuzzy map on a simplified version of `Type`. -- Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9805> GHC <http://www.haskell.org/ghc/> The Glasgow Haskell Compiler

1 17