Summary: 364 instances, 353 unique

Text	Count
-- TODO what else to count while we're here?	1
-- it is a reg.  FIXME: remove this restriction.	2
-- spectral/puzzle. TODO Perhaps adjusting the default threshold would be a	1
-- /TODO:/ Add usage example?	1
return "" -- TODO: Use a safe interface.	1
-- TODO: this is a hack.	1
-- TODO: do we need to support `includes_CC_OPTS += -DDYNAMIC_BY_DEFAULT`?	1
-- TODO: We cannot retry on EINTR as the timeout would be wrong.	1
-- TODO: Add some sanity checking!!!!	1
= let !q' = plusInteger q (S# -1#) -- TODO: optimize	1
-- TODO : Should we remove the previosly generated config file?	1
-- TODO This approach doesn't work.	1
// FIXME again: perhaps we should throw a synchronous exception	1
-- TODO Check that the above notes on @f@ are valid. The function successfully	1
* TODO: measure whether we really need these, or whether Mutexes	1
-- TODO: This should be fixed properly, e.g. generated here on demand.	1
newUnitId cid [] = newSimpleUnitId cid -- TODO: this indicates some latent bug...	1
-- TODO: If a wired in package had a compiler plugin,	1
ms_obj_date = Nothing, -- TODO do this, but problem: hi_timestamp is BOGUS	1
-- TODO: Several primops, such as 'copyArray#', only have an inline	1
-- TODO: This could be very well generalized to support instance	1
-- TODO: toIfaceTcArgs seems rather wasteful here	1
-- Explicit definition to avoid dependency on Data.List.Ordered. TODO: add tests	1
-- TODO: is LL right here?	1
-- TODO are we still tracking "good stuff" (_gds) versus	1
-- TODO: We could also check for the last asm instruction which sets	1
-- TODO: need to special case wired in names	1
-- TODO: Move @timeout@ to the @util@ directory and build in a more standard	1
-- TODO recomp014 triggers this assert. What's going on?!	1
-- FIXME: do we really need to zero-init MBAs returned by 'newByteArray#'?	1
-- TODO: an alternative would be to generalize by linking with extra	1
-- TODO: below we use it over temporary register,	1
-- TODO: we could save a byte in a subsequent CMP instruction too,	1
-- TODO: Switch to 'Stage' as the first argument instead of 'String'.	2
-- TODO: Figure out why :reload doesn't load the docs and maybe fix it.	1
// TODO: need to trace various other things in the child	1
-- TODO: get rid of non-derived Show instances	1
-- TODO: There are other interesting patterns we want to replace	1
/* TODO: longlong.c */	1
-- TODO: Maybe setGlobalTypeEnv should be strict.	1
-- TODO: Since we JUST deprecated	1
TODO: Depending on how much allocation overhead stgMallocBytes uses for	1
-- TODO: Need compiler_stage1_CONFIGURE_OPTS += --disable-library-for-ghci?	1
ArchPPC           -> 64 -- TODO: check ABI spec	1
-- TODO: Optimize branch for common case of no aliasing.	3
// TODO: consider implementing this function in Haskell-land	2
-- TODO: Allow us to distinguish between abstract class,	1
-- TODO:	2
let myprec = 10 -- application precedence. TODO Infix constructors	1
* TODO: Consider to switch return type from 'uint32_t' to 'StgWord' #8742	1
/* TODO: Only if !full GC */	1
= let -- TODO: Literal labels might not actually be in the current package...	1
-- TODO revisit this if we ever do boxity analysis	1
-- TODO: Make this have the proper information	1
-- TODO: if saturated arithmetic is to used, replace 'underflowError' by '0'	1
-- TODO: Is this still needed? See 'runtimeDependencies'.	1
--      TODO:   It'd be better to use a scratch reg instead of the add/sub thing.	1
#define BlockedOnMVarRead   14 /* TODO: renumber me, see #9003 */	1
dataTcOccs, --TODO: Move this somewhere, into utils?	1
// TODO: this may waste some work, if the thunk has	1
-- TODO: maybe super classes should be allowed in abstract class definitions	1
-- TODO: Should @PackageConfiguration@ be simply @()@? Presumably the pair	1
GarbageCollect can check.  (not ideal, TODO: better)	1
-- TODO: Reduce the variety of similar flags (e.g. CPP and non-CPP versions).	1
-- TODO (RAE): Change back to ASSERT	1
-- TODO: We should do a no_change check for the dynamic	1
-- TODO: change loadDecls to accept w/o Fingerprint	1
-- TODO: Optimise by switching to sets of packages.	1
-- TODO: Pass the correct way from Rules via Context.	1
-- TODO: This should not be @not <$> flag CrossCompiling@. Instead we	1
# TODO: Remove this once it's no longer necessary	1
-- TODO: What about docs for constructors etc.?	1
-- TODO: Find a way to differentiate between modules loaded without '-haddock'	1
-- TODO: A real dataflow analysis would do a better job here.	1
-- FIXME: it is possible that Handle code using the haDecoder/haEncoder fields	1
-- TODO:	1
-- TODO shouldn't ignore result:	1
-- TODO: switch to Set Package as the order of packages should not matter?	1
-- TODO: Get rid of deprecated 'externalPackageDeps' and drop -Wno-deprecations	1
$white_no_nl = $whitechar # \n -- TODO #8424	1
GT -> case runS go of (!q,!(W# r#)) -> (# q, r# #) -- TODO: handle word/word	1
-- TODO: Add movabs instruction and support 64-bit sets.	1
-- TODO: Not sure if it's worth to write 'Float' optimized versions here	1
TODO: If we remove edges as we use them (eg if we build up A->B remove A->B	1
-- TODO: This used to include HaskellConstants.hs, but that has now gone.	1
-- XXX: TODO: PRETTY: Used to use True here (but GHC used False...)	2
TODO: Currently this code does not tag created pointers,	1
TODO:	1
sqrBigNat x = timesBigNat x x -- TODO: mpn_sqr	1
* TODO: Merge SymbolExtras and Stubs.	1
--TODO: Partially check validity of the cfg.	1
-- TODO: This is copied from the ReadOnlyData case, but it can likely be	1
--      TODO: We should also make arch specific versions of RegAlloc.Graph.TrivColorable	1
/* TODO: implement w/o GMP, c.f. 'integer_gmp_mpn_import()' */	1
-- TODO: In that case we also do not need to build most of the Stage1	1
-- it is currently missing. TODO: Is this the right thing to do?	1
-- /TODO:/ Also clear up confusion about 'Rec0' and 'Rec1' not really indicating recursion.	1
-- FIXME: Implement and provide instances for Eq and Storable	1
-- FIXME: this is not a complete solution because if the locale encoding is one	1
-- TODO: Do not create a new stack frame if delta is too large.	1
// TODO: Remove this code path, it cannot be taken because	1
-- Helpers for 'Enum Natural'; TODO: optimise & make fusion work	1
TODO: Encode this via an extension to GenStgExpr à la TTG.	1
-- TODO: Investigate if these help or hurt when using split sections.	1
-- TODO: work this into GhcMake!!	1
-- TODO: should `elem` be `wayUnit`?	1
-- TODO: Preprocessing not implemented	1
let loc = mkRealSrcLoc (mkFastString src_filename) 1 1 -- TODO: not great	1
| otherwise -- No match, so push back (FIXME O(n^2))	1
-- TODO should this be somewhere else?	1
--   TODO: See if we can split some of the live ranges instead of just globally	1
-- TODO: See https://github.com/snowleopard/hadrian/issues/549.	1
This appears to be very rare in practice. TODO Perhaps we should gather	1
-- TODO: show annotations here	1
-- constraints. TODO: Do the same for the type constraints?	1
-- TODO: clean up after testing	1
("Run " ++ show (builder t)) -- TODO: Bring back contextInfo.	2
// TODO: can we remove this?	1
-- TODO: add some checks here	1
-- (TODO: I don't think we need this --SDM)	1
-- TODO: Use optimization fuel	1
FIXME Currently the functionality to produce "eta-contracted" wrappers is	1
-- TODO: Actually this error swizzle doesn't work	1
-- TODO: Should OutputableX be included in OutputableBndrId?	1
TODO: Make sure that kicking out really *is* a Bad Thing. We've assumed	1
classMinimalDef _ = mkTrue -- TODO: make sure this is the right direction	1
-- TODO : Use input test compiler and not just stage2 compiler.	1
-- TODO: Find a more direct indicator.	1
extra_src_files = {'T4198': ['exitminus1.c']} # TODO: See #12223	1
hpm_src_files = [], -- TODO if we preprocessed it	1
TODO:AZ: remove this, and check if we still need all the UndecidableInstances	1
-- TODO: also print arg docs.	1
-- TODO is it ever called with a Single-Entry thunk?	1
-- TODO: this specializing is not allowed	1
hiddenModules = [], -- TODO: doc only	1
-- better errors.  (TODO: how?)	1
-- TODO: experiment with different constructor-ordering	1
-- FIXME: share the buffer between runs, even though that means we can't size the buffer as we want.	1
-- TODO: For small types look at the tag bits instead of reading info table	1
-- TODO: maybe we can be a little more	1
//TODO: we ignore local symbols anyway right? So we can use the	1
// FIXME: do some more verifying here	1
_        -> do -- TODO: Load from TOC,	1
-- FIXME: we should also warn about view patterns that should be	1
-- TODO: do we also have to check the trust of the instantiation?	1
-- TODO: Revisit this	1
--TODO: Remove - Just for development/debugging	1
--TODO: Update CFG for x86	1
// TODO: fill info	1
-- FIXME: Why?  The actual comparison is not hard, but what would	1
-- TODO: Eliminate code duplication between the checked and unchecked	1
-- TODO: This is fragile: we should remove this from behind the	1
* TODO: check if this comment is still relevant.	1
// FIXME this is not a good way to tell a program to release	1
-- TODO: replace 'Int' by an enum such as 'PPC_64ABI'	1
-- | TODO: Drop code duplication	1
-- TODO: Maybe also get module headers for module names	1
-- TODO: try to get rid of this instance	1
-} -- TODO Printing infix constructors properly	1
-- TODO: there's no need for these to be FastString, as we don't need the uniq	1
-- TODO: Make sure this error message is good	1
= pprPanic "TODO: NPlusKPat" $ ppr n $$ ppr k $$ ppr geq $$ ppr subtract	1
-- TODO: Make this list complete.	1
--      TODO: Is that still true? Could we use allocatableRegsInClass	1
-- TODO Check if we actually *are* changing any join points' types	1
-- TODO: The @touchy@ utility lives in the @lib/bin@ directory instead of	1
-- TODO: Maybe lcl_iface0 should be pre-renamed to the right thing? Unclear...	1
ppr (L l e) = -- TODO: We can't do this since Located was refactored into	1
-- TODO: Combine this with 'programName'.	1
matching (UnitIdArg _)  = \_ -> False -- TODO: warn in this case	1
-- TODO: Change @mk/config.h@ to @shake-build/cfg/config.h@.	1
-- TODO: Get rid of the @includeGhciLib@ hack.	1
-- TODO: maybe associate this with a source location?	1
-- TODO:AZ these constraints do not make sense	1
-- TODO There may be ways to make ticks and join points play	1
-- TODO: Can we get rid of this enumeration of contexts? Since we iterate	1
// TODO: asserting task->cap == NULL would be much stronger	1
-- TODO: Several primop implementations (e.g. 'doNewByteArrayOp') use	1
-- TODO: Isn't vanilla always built? If yes, some conditions are redundant.	1
// TODO: fix this story proper, so that the transformation	1
/* TODO: Stop relying on file extensions to determine input formats.	1
-- TODO: could gather equalities from superclasses too	1
--         of an object file.  (TODO: do we actually need this?)	1
-- TODO: This code is duplicated in compiler/main/Packages.hs	1
-- TODO: I don't know how to get this right for rebindable syntax	1
-- TODO: this should not be necessary	1
-- TODO: Augment this Makefile to match the various parameters that the current	1
--      TODO:   add work lists to finding triv nodes is easier.	1
// TODO: This sends multiple messages if we write to the same TVar multiple	1
-- TODO: we should do a non-blocking flush here	1
-- otherwise Windows breaks. TODO: Figure out why.	1
-- TODO: The use of nameOccName here is a bit dodgy, because	1
-- TODO: This is wrong	1
-- TODO: Get rid of this hack.	1
# TODO (int-index): restore the -c option when happy/pull/134 is merged.	1
-- TODO: Track command line arguments and package configuration flags.	1
{- TODO: reserve r13 on 64 bit systems only and r30 on 32 bit respectively.	1
-- TODO: Cache the computation.	1
-- FIXME: some of these panics should be converted into proper error messages	1
// TODO I'm not seeing where ALLOC_P_TICKY is used; can it be removed?	1
TODO: The inability to handle objects bigger than BLOCK_SIZE_W means that	1
-- TODO: Should handle the dynamic hi filename properly	1
-- FIXME: I haven't got the types here right yet	1
getLit _ = panic "invalid literal" -- TODO messy failure	1
-- TODO: When it's a synonym implementing a class, we really	1
-- TODO: This will give an unpleasant message if n'	1
-- TODO: Questionable.	1
--      TODO:   Signed vs unsigned?	1
-- TODO: Figure out what the correct form of this assert is. It's violated	1
-- TODO: make this the only type for ValBinds	1
{- TODO: enable this later, when we have some good tests in place to	1
--  TODO: add support for prefetch WRITES,	1
lambda without introducing a type let requires a substitution). TODO: Improve	1
-- TODO: This code is duplicated in utils/ghc-pkg/Main.hs	1
-- TODO: setup the local RdrEnv so the error messages look a little better.	1
-- TODO: The 'Semigroup a' constraint will at some point become redundant.	1
-- TODO: ghc -M is unlikely to do the right thing	1
--   TODO: I'm not entirely sure if this is catching what we really want to catch	1
--FIXME: we don't really want to show the unit id here we should	1
/* TODO: Don't just take first block, find smallest sufficient block */	1
// TODO: do we need to traceTask*() here?	1
-- TODO: Actually this should never happen, because DFuns are	1
-- TODO: align closures only	1
-- TODO: just make a proper new monad for BkpM, rather than use IOEnv	1
-- TODO: Why is this ok? Under linux this code will be in libm.so	1
-- See Note [Unit tuples] in HsTypes  (TODO: is this still relevant?)	1
-- TODO: Make PackageType more precise.	1
-- TODO: I'm not sure if this should be semantic_mod or this_mod.	1
--      TODO:   This module is currenly shared by all architectures because	1
-- tcg_env (TODO: but maybe this isn't relevant anymore).	1
-- TODO: there is duplication and inconsistency between this and	1
-- TODO: we may be able to replace the following with utils from the binary	1
, "# TODO : find if a better function exists"	1
cons_prec = 5 -- TODO Extract this info from GHC itself	1
(tcg, mod_guts) <- -- TODO: space leaky: call hsc* directly?	1
-- TODO: WRITE MORE DETAILS ON HOW THIS WORKS.	1
-- TODO: stop repeatedly rebuilding the OccEnv	1
--      TODO: Why can we assume that o6 isn't clobbered? -- BL 2009/02	1
It's not clear that this comes up often, however. TODO: Measure how often and	1
-- TODO: use Sum monoid	1
-- TODO: Make this type more precise	1
if not windows && -- TODO: We don't use system GMP on Windows. Fix?	1
-- TODO: make this check a function	1
; -- TODO This is a little skeevy; maybe handle a bit more directly	1
them as givens when checking the RHS. TODO (RAE): Implement plan.	1
-- TODO: Ticks in unfoldings. Maybe change unfolding so it removes	1
-- TODO: Consider turning Package into a GADT indexed with language and type.	1
-- TODO: Explain this better and refer to ABI spec!	1
ctx <- programContext stage pkg -- TODO: see todo on programContext.	1
-- TODO: (S# -1) cases	1
-- TODO: We need to set @--with-strip=(stripCmdPath :: Action FilePath)@,	1
-- TODO: Distinguish between signature merging and signature	1
--      TODO: Do a better job of the overloading, and eliminate this module.	1
-- TODO: remove this use of append somehow (use a data structure with O(1) append but a left-to-right kind of interface)	1
(Used only for PArray, TODO: remove?)	1
-- TODO: check code duplication w.r.t. ConfCcArgs	1
-- TODO: This is conceptually wrong!	1
-- TODO: Get rid of this hack.	2
-- TODO: Move this to build directory?	1
--      TODO: It might be nice to extend these constructors with information	1
-- TODO: Can we extract this information from Cabal files?	1
-- TODO: should be in a different file	1
-- TODO: we should do a non-blocking flush here	1
-- TODO: Maybe move this to HsExtensions? I'm not sure about the implications	1
-- TODO Consider whether this is the correct rule. It is consistent with	1
--      TODO:   When sign extending, GCC only shifts the a_reg right by 17 bits,	1
-- TODO (DV): these are only really used in the	1
-- TODO: Speed up? Switch to Set?	1
-- TODO: move inside withLibLBI	1
-- TODO: Do we really want to the extension point type families to have a closed	1
-- TODO: Make this setting configurable? Setting to True by default seems	1
-- import qualified Data.IntMap.Strict as M --TODO: LabelMap	1
pc -= 1; // TODO: is this right?	1
-- TODO undecidable instance	1
-- TODO: Location of 'gmpBuildPath' is important: it should be outside any	1
-- TODO  This instance is not architecture portable.  GMP stores numbers as	1
-- TODO (RAE): This is inadequate, as it doesn't flatten the kind of	1
{-  | dataConIsInfix dc, (t1:t2:tt') <- tt  --TODO fixity	1
-- TODO could optimize with an aborting variant of zipWith since	1
-- TODO: This is copied from the ReadOnlyData case, but it can likely be	1
-- TODO: Possible optimisation:	1
-- TODO: Add levity polymorphism check for the resulting expression.	1
-- TODO: Windows supports lazy binding (but GHC doesn't currently support	1
-- TODO: This should really just use GHC's Unique and Uniq{Set,FM}	1
-- TODO: 1-limb BigNats below sqrt(maxBound::GmpLimb)	1
= c_mpn_export1ToAddr# w# addr 0# -- TODO: we don't calling GMP for that	1
-- TODO: simplify, get rid of fake rts context	1
-- TODO: Do we really need all these modes? Why do we need 'Dependencies'? We	1
-- TODO what about windows?	1
happyBuilderArgs = builder Happy ? mconcat [ arg "-ag" -- TODO (int-index): restore the -c option when happy/pull/134 is merged.	1
let ghcBuiltByLlvm = False -- TODO: Implement this check	1
, arg "-e", arg $ "ghc_with_llvm=0" -- TODO: support LLVM	1
-- TODO: remove the redundant case in the original code.	1
-- | TODO: Is there a simpler way to find parent directory?	1
|     -- powerPC NCG has a TODO for I8/I16 comparisons, so don't try	1
-- TODO: perhaps we shouldn't do this if there are any strict bindings,	1
TODO: Should these counters be more or less granular?	1
-- TODO: LD_OPTS?	1
-- TODO: Do '-ticky' in all debug ways?	1
-- | TODO: Move to Hadrian utilities.	1
-- TODO: Warnings are transitive, but this is not what we want here:	1
-- TODO (SM): benchmark this against a tail-recursive variant	1
-- TODO: More commentary needed here	1
-- TODO: Shall we use Stage2 for testsuite packages instead?	1
-- TODO: would love to have context	1
// TODO:	2
-- TODO: Use the Types, Luke! (drop partial function)	1
-- TODO: locs are wrong	1
-- TODO (Alp): we might sometimes need more than vanilla!	1
-- TODO: Note that some functions have commented CONSTANT_FOLDED annotations,	1
-- TODO: this seems a bit dodgy, see 'eqTickish'	1
-- TODO: reconcile with includes/Regs.h	1
-- | Update the EPS from a 'GhcMonad'. TODO move to appropriate library spot.	1
-- TODO: Can we extract this information from Cabal files?	1
* TODO: find a better way to do this	1
TODO: a step-by-step replay of the refactor to analyze the performance.	1
-- TODO: this looks for modules in the wrong place	1
-- TODO: Move C source arguments here	1
-- TODO: Do we still need to clean the tree?	1
-- TODO If this function is performance-critical, it would probably	1
// TODO: over/underflow handling?	1
-- TODO (SM): get rid of bang patterns	1
-- FIXME: share the buffer between runs, even if the buffer is not the perfect size	1
-- TODO: generate noReloads from liveSlotsOnEntry	1
TODO: This can probably be solved somehow in SetLevels. The difference between	1
TODO: make sure this still works, it might be bitrotted	1
-- TODO (can't be bothered right now)	1
/* TODO: implement w/o GMP */	1
-- TODO: set CABAL_MINIMAL_BUILD/CABAL_PLUGIN_BUILD	1
-- TODO: refactor	1
-- TODO: Remove Fun when #14253 is fixed	1
-- TODO/FIXME promotion-hack	2
-- TODO: Cache this, so that we don't repeatedly re-check	1
-- TODO: rename plzTick => plzExprTick, plzTopPick => plzTick	1
-- TODO: Remove stable pointer hack after fixing #9718.	1
-- TODO: won't work if home dir is encoded.	1
-- TODO: This is mostly inter-block	1
-- TODO: The use case here is that we want to build @ghc-proxy@ for the	1
-- TODO: Remove this hack.	1
--      TODO (osa): Why?	1
-- AZ:TODO: Change these all to be Name instead of RdrName.	1
-- TODO: include windows_cconv.h when it's merged, instead of duplicating	1
-- TODO: Instead of using a separate predicate (i.e. isSse2Enabled) to	1
-- TODO: the regClass function is currently hard coded to the default	1
-- AZ:TODO above comment needs to be updated	1
--   TODO: On CISCy x86 and x86_64 we don't necessarily have to add a mov instruction	1
The dictionary gets type [C * (D *)]. IA0_TODO it should be	1
-- AZ:TODO: use XSplice instead of HsSpliced	1
-- TODO: Some builders are required only on certain platforms. For example,	1
-- TODO: either should be Module, or there	1
# TODO: Add {Open,Free}BSD support	1
// TODO: This seems wrong to me. Surely we should be subtracting	1
-- TODO: See Note [Excess polymorphism and join points]	1