Commit 11390e3a authored by Takenobu Tani's avatar Takenobu Tani Committed by Marge Bot
Browse files

Clean up file paths for new module hierarchy

This updates comments only.
This patch replaces file references according to new module hierarchy.

See also:
* https://gitlab.haskell.org/ghc/ghc/-/wikis/Make-GHC-codebase-more-modular
* ghc/ghc#13009
parent 7002d0cb
......@@ -1346,7 +1346,7 @@ getNameToInstancesIndex visible_mods mods_to_load = do
-- -----------------------------------------------------------------------------
{- ToDo: Move the primary logic here to compiler/main/Packages.hs
{- ToDo: Move the primary logic here to compiler/GHC/Unit/State.hs
-- | Return all /external/ modules available in the package database.
-- Modules from the current session (i.e., from the 'HomePackageTable') are
-- not included. This includes module names which are reexported by packages.
......
......@@ -542,7 +542,7 @@ toBlockListEntryFirst g
-- have both true and false successors. Block ordering can make a big difference
-- in performance in the LLVM backend. Note that we rely crucially on the order
-- of successors returned for CmmCondBranch by the NonLocal instance for CmmNode
-- defined in cmm/CmmNode.hs. -GBM
-- defined in GHC/Cmm/Node.hs. -GBM
toBlockListEntryFirstFalseFallthrough :: CmmGraph -> [CmmBlock]
toBlockListEntryFirstFalseFallthrough g
| mapNull m = []
......
......@@ -3,7 +3,7 @@
-- Also import the required constants, so we know what we're using.
--
-- In the interests of cross-compilation, we want to free ourselves
-- from the autoconf generated modules like main/Constants
-- from the autoconf generated modules like GHC/Settings/Constants
module GHC.CmmToAsm.SPARC.Base (
wordLength,
......
......@@ -866,7 +866,7 @@ False) and that all is OK. But, all is not OK: we want to use the first branch
of the axiom in this case, not the second. The problem is that the parameters
of the first branch can unify with the supplied coercions, thus meaning that
the first branch should be taken. See also Note [Apartness] in
types/FamInstEnv.hs.
GHC/Core/FamInstEnv.hs.
Note [Why call checkAxInstCo during optimisation]
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
......
......@@ -674,7 +674,7 @@ in this case by a little known `optimization' that was disabled in
an instance declaration. In this case, it silently inserts the `C
[a]', and everything happens to work out.
(See `basicTypes/MkId:mkDictFunId' for the code in question. Search for
(See `GHC/Types/Id/Make:mkDictFunId' for the code in question. Search for
`Mark Jones', although Mark claims no credit for the `optimization' in
question, and would rather it stopped being called the `Mark Jones
optimization' ;-)
......
......@@ -1552,7 +1552,7 @@ match_inline (Type _ : e : _)
match_inline _ = Nothing
-- See Note [magicDictId magic] in `basicTypes/MkId.hs`
-- See Note [magicDictId magic] in `GHC/Types/Id/Make.hs`
-- for a description of what is going on here.
match_magicDict :: [Expr CoreBndr] -> Maybe (Expr CoreBndr)
match_magicDict [Type _, Var wrap `App` Type a `App` Type _ `App` f, x, y ]
......
......@@ -146,7 +146,7 @@ tidyType env (CoercionTy co) = CoercionTy $! (tidyCo env co)
-- The following two functions differ from mkForAllTys and splitForAllTys in that
-- they expect/preserve the ArgFlag argument. These belong to types/Type.hs, but
-- they expect/preserve the ArgFlag argument. These belong to GHC/Core/Type.hs, but
-- how should they be named?
mkForAllTys' :: [(TyCoVar, ArgFlag)] -> Type -> Type
mkForAllTys' tvvs ty = foldr strictMkForAllTy ty tvvs
......
......@@ -1372,7 +1372,7 @@ although it's probably true that we could merge some of these.
Roughly in order of "includes more information":
- A Width (cmm/CmmType) is simply a binary value with the specified
- A Width (GHC/Cmm/Type) is simply a binary value with the specified
number of bits. It may represent a signed or unsigned integer, a
floating-point value, or an address.
......
......@@ -991,7 +991,7 @@ binary sizes shrink significantly either.
Note [Discounts and thresholds]
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Constants for discounts and thesholds are defined in main/DynFlags,
Constants for discounts and thesholds are defined in GHC/Driver/Session,
all of form ufXxxx. They are:
ufCreationThreshold
......
......@@ -1276,7 +1276,7 @@ for all other cases, apart from EmptyCase). This gave rise to #10746. Instead,
we do the following:
1. We normalise the outermost type family redex, data family redex or newtype,
using pmTopNormaliseType (in types/FamInstEnv.hs). This computes 3
using pmTopNormaliseType (in GHC/Core/FamInstEnv.hs). This computes 3
things:
(a) A normalised type src_ty, which is equal to the type of the scrutinee in
source Haskell (does not normalise newtypes or data families)
......@@ -1291,7 +1291,7 @@ we do the following:
newtype rewrite performed in (b).
For an example see also Note [Type normalisation]
in types/FamInstEnv.hs.
in GHC/Core/FamInstEnv.hs.
2. Function Check.checkEmptyCase' performs the check:
- If core_ty is not an algebraic type, then we cannot check for
......
......@@ -271,7 +271,7 @@ The impact of this treatment of overloaded literals is the following:
* We have instant equality check for overloaded literals (we do not rely on
the term oracle which is rather expensive, both in terms of performance and
memory). This significantly improves the performance of functions `covered`
`uncovered` and `divergent` in deSugar/Check.hs and effectively addresses
`uncovered` and `divergent` in GHC/HsToCore/PmCheck.hs and effectively addresses
#11161.
* The warnings issued are simpler.
......
......@@ -200,7 +200,7 @@ call and just recurse directly in to the subexpressions.
-}
-- These synonyms match those defined in main/GHC.hs
-- These synonyms match those defined in compiler/GHC.hs
type RenamedSource = ( HsGroup GhcRn, [LImportDecl GhcRn]
, Maybe [(LIE GhcRn, Avails)]
, Maybe LHsDocString )
......
......@@ -187,7 +187,7 @@ mkIfaceForAllTvBndr :: ArgFlag -> IfaceTvBndr -> IfaceForAllBndr
mkIfaceForAllTvBndr vis var = Bndr (IfaceTvBndr var) vis
-- | Build the 'tyConKind' from the binders and the result kind.
-- Keep in sync with 'mkTyConKind' in types/TyCon.
-- Keep in sync with 'mkTyConKind' in GHC/Core/TyCon.
mkIfaceTyConKind :: [IfaceTyConBinder] -> IfaceKind -> IfaceKind
mkIfaceTyConKind bndrs res_kind = foldr mk res_kind bndrs
where
......
......@@ -119,7 +119,7 @@ import GHC.Parser.Annotation
-- -----------------------------------------------------------------------------
-- Alex "Character set macros"
-- NB: The logic behind these definitions is also reflected in basicTypes/Lexeme.hs
-- NB: The logic behind these definitions is also reflected in GHC/Utils/Lexeme.hs
-- Any changes here should likely be reflected there.
$unispace = \x05 -- Trick Alex into handling Unicode. See [Unicode in Alex].
$nl = [\n\r\f]
......@@ -2261,7 +2261,7 @@ adjustChar c = fromIntegral $ ord adj_c
-- with the actual character value hidden in the state.
| otherwise =
-- NB: The logic behind these definitions is also reflected
-- in basicTypes/Lexeme.hs
-- in GHC/Utils/Lexeme.hs
-- Any changes here should likely be reflected there.
case generalCategory c of
......
......@@ -166,7 +166,7 @@ Other Notes on Remote GHCi
~~~~~~~~~~~~~~~~~~~~~~~~~~
* This wiki page has an implementation overview:
https://gitlab.haskell.org/ghc/ghc/wikis/commentary/compiler/external-interpreter
* Note [External GHCi pointers] in compiler/ghci/GHCi.hs
* Note [External GHCi pointers] in compiler/GHC/Runtime/Interpreter.hs
* Note [Remote Template Haskell] in libraries/ghci/GHCi/TH.hs
-}
......
......@@ -902,7 +902,7 @@ cgIdApp fun_id args = do
--
-- Self-recursive tail calls can be optimized into a local jump in the same
-- way as let-no-escape bindings (see Note [What is a non-escaping let] in
-- stgSyn/CoreToStg.hs). Consider this:
-- GHC/CoreToStg.hs). Consider this:
--
-- foo.info:
-- a = R1 // calling convention
......
......@@ -241,7 +241,7 @@ emitRtsCallGen res lbl args safe
-- here, as we don't have liveness information. And really, we
-- shouldn't be doing the workaround at this point in the pipeline, see
-- Note [Register parameter passing] and the ToDo on CmmCall in
-- cmm/CmmNode.hs. Right now the workaround is to avoid inlining across
-- GHC/Cmm/Node.hs. Right now the workaround is to avoid inlining across
-- unsafe foreign calls in rewriteAssignments, but this is strictly
-- temporary.
callerSaveVolatileRegs :: DynFlags -> (CmmAGraph, CmmAGraph)
......
......@@ -155,7 +155,7 @@ mkNoteObjsToLinkIntoBinary dflags dep_packages = do
-- ALL generated assembly must have this section to disable
-- executable stacks. See also
-- compiler/nativeGen/AsmCodeGen.hs for another instance
-- compiler/GHC/CmmToAsm.hs for another instance
-- where we need to do this.
if platformHasGnuNonexecStack platform
then text ".section .note.GNU-stack,\"\","
......
......@@ -452,7 +452,7 @@ makeDerivSpecs deriv_infos deriv_decls
-- | Process the derived classes in a single @deriving@ clause.
deriveClause :: TyCon
-> [(Name, TcTyVar)] -- Scoped type variables taken from tcTyConScopedTyVars
-- See Note [Scoped tyvars in a TcTyCon] in types/TyCon
-- See Note [Scoped tyvars in a TcTyCon] in GHC/Core/TyCon
-> Maybe (LDerivStrategy GhcRn)
-> [LHsSigType GhcRn] -> SDoc
-> TcM [EarlyDerivSpec]
......
......@@ -2041,7 +2041,7 @@ specially in `GHC.HsToCore.Quote`, `GHC.ThToHs`, and
(where ty is the AST representation of t1 -> t2 -> ... -> tn -> t)
(b) When converting pattern synonyms from TH.Dec to HsSyn in
`hsSyn/Convert.hs`, we convert their TH type signatures back to an
`GHC/ThToHs.hs`, we convert their TH type signatures back to an
appropriate Haskell pattern synonym type of the form
forall univs. reqs => forall exis. provs => t1 -> t2 -> ... -> tn -> t
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment