[commit: ghc] master: Efficient membership for home modules (26f509a)

[git at git.haskell.org]

Repository : ssh://git@git.haskell.org/ghc

On branch  : master
Link       : http://ghc.haskell.org/trac/ghc/changeset/26f509a992ebc6910ed2309b46f3f1d44efba7c9/ghc

>---------------------------------------------------------------

commit 26f509a992ebc6910ed2309b46f3f1d44efba7c9
Author: Bartosz Nitka <niteria at gmail.com>
Date:   Wed May 10 04:36:52 2017 -0700

    Efficient membership for home modules
    
    This changes the linear lookup in a list to an efficient
    lookup in an IntMap. The linear lookup effectively made
    the algorithm quadratic, which for a test case that I have
    (5000 modules) introduced significant slowdown.
    
    I ran 3 experiments to estimate the impact of this:
    
    "No-op", profiled, just `:load`: P146, `186s`
    "before", profiled, `:load` followed by 10x `:r`: P147, `315s`
    "after", profiled, `:load` followed by 10x `:r`: P148, `250s`
    
    Going by the math of `(250-186)/(315-186) = 50%` this is a 2x improvement
    on `:r`.
    
    Test Plan: ./validate
    
    Reviewers: simonmar, austin, bgamari
    
    Reviewed By: simonmar
    
    Subscribers: rwbarton, thomie
    
    Differential Revision: https://phabricator.haskell.org/D3562


>---------------------------------------------------------------

26f509a992ebc6910ed2309b46f3f1d44efba7c9
 compiler/main/GhcMake.hs | 12 +++++++-----
 1 file changed, 7 insertions(+), 5 deletions(-)

diff --git a/compiler/main/GhcMake.hs b/compiler/main/GhcMake.hs
index 7cc5276..4d06b6e 100644
--- a/compiler/main/GhcMake.hs
+++ b/compiler/main/GhcMake.hs
@@ -220,8 +220,9 @@ load' how_much mHscMessage mod_graph = do
     -- B.hs-boot in the module graph, but no B.hs
     -- The downsweep should have ensured this does not happen
     -- (see msDeps)
-    let all_home_mods = [ms_mod_name s
-                        | s <- mod_graph, not (isBootSummary s)]
+    let all_home_mods =
+          mkUniqSet [ ms_mod_name s
+                    | s <- mod_graph, not (isBootSummary s)]
     -- TODO: Figure out what the correct form of this assert is. It's violated
     -- when you have HsBootMerge nodes in the graph: then you'll have hs-boot
     -- files without corresponding hs files.
@@ -236,7 +237,7 @@ load' how_much mHscMessage mod_graph = do
         checkHowMuch _ = id
 
         checkMod m and_then
-            | m `elem` all_home_mods = and_then
+            | m `elementOfUniqSet` all_home_mods = and_then
             | otherwise = do
                     liftIO $ errorMsg dflags (text "no such module:" <+>
                                      quotes (ppr m))
@@ -656,7 +657,7 @@ unload hsc_env stable_linkables -- Unload everthing *except* 'stable_linkables'
 checkStability
         :: HomePackageTable   -- HPT from last compilation
         -> [SCC ModSummary]   -- current module graph (cyclic)
-        -> [ModuleName]       -- all home modules
+        -> UniqSet ModuleName -- all home modules
         -> ([ModuleName],     -- stableObject
             [ModuleName])     -- stableBCO
 
@@ -669,7 +670,8 @@ checkStability hpt sccs all_home_mods = foldl checkSCC ([],[]) sccs
      where
         scc = flattenSCC scc0
         scc_mods = map ms_mod_name scc
-        home_module m   = m `elem` all_home_mods && m `notElem` scc_mods
+        home_module m =
+          m `elementOfUniqSet` all_home_mods && m `notElem` scc_mods
 
         scc_allimps = nub (filter home_module (concatMap ms_home_allimps scc))
             -- all imports outside the current SCC, but in the home pkg