-cvs remove TcGRHSs.hi-boot TcGRHSs.hi-boot-5 TcGRHSs.lhs
-cvs remove pbinding.ugn
-cvs add grhsb.ugn gdexp.ugn
------------------------
+ New back end thoughts
+-----------------------------------------------------------------------------
+Codegen notes
-* CHECK that the things seek_liftable found are done in Core
+* jumps to ImpossibleBranch should be removed.
-* CHECK that there aren't too many indirections in STG
- local = ...
- global = local Int
+* Profiling:
+ - when updating a closure with an indirection to a function,
+ we should make a permanent indirection.
-Interface files
-~~~~~~~~~~~~~~~
-* Don't need to pin a kind on the type variable in a interface class decl,
- because it'll be correctly re-inferred when we read it in.
+ - check that we're bumping the scc count appropriately
-* The double semicolon at the end of an interface-file signature is so that
- the lexer can run through the pragmas very fast when -O isn't being used.
+* check perf & binary sizes against the HEAD
-* In export lists, T|(A,B) says that constructors A and B are exported,
- but not the type T. Similarly for classes.
- We can't say T(T,A,B) and T(A,B) to export or not-export T respectively,
- because the type T might have a constructor T.
+-----------------------------------------------------------------------------
+C backend notes
-===========================================================================
+* use STGCALL macros for foreign calls (doesn't look like volatile regs
+ are handled properly at the mo).
- Nofib failures
- ~~~~~~~~~~~~~~
+-----------------------------------------------------------------------------
+Cmm parser notes
-* spectral/hartel/wave4main, wang, spectral/simple, real/symalg
+* switches
-Bus error
+* need to cater for unexported procedures/info tables?
-* real/anna
+* We should be able to get rid of entry labels, use info labels only.
+ - we need a %ENTRY_LBL(info_lbl) macro, so that instead of
+ JMP_(foo_entry) we can write jump %ENTRY_LBL(foo_info).
-expected stdout not matched by reality
-*** big.sum.out Thu Aug 22 14:37:05 1996
---- /tmp/runtest21900.1 Mon Jan 20 17:57:49 1997
-***************
-*** 1 ****
-! 12796 49
---- 1 ----
-! 63325 97
+-----------------------------------------------------------------------------
+* Move arg-descr from LFInfo to ClosureInfo?
+ But: only needed for functions
-* /real/compress2
+* Move all of CgClosure.link_caf into NewCaf, and newDynCaf
-expected stderr not matched by reality
-Warning: missing newline at end of file /tmp/runtest14691.2
-*** /tmp/no_stderr14691 Thu Jan 23 14:33:29 1997
---- /tmp/runtest14691.2 Thu Jan 23 14:33:29 1997
-***************
-*** 0 ****
---- 1,2 ----
-+
-+ Fail: Prelude.Enum.Char.toEnum:out of range
+* If the case binder is dead, and the constr is nullary,
+ do we need to assign to Node?
-* real/ebnf2ps
-
-IOSupplement.hs: 43: value not in scope: getEnv
+-------------------------
+* Relation between separate type sigs and pattern type sigs
+f :: forall a. a->a
+f :: b->b = e -- No: monomorphic
- ...and...
-
-HappyParser.hs: 127: Couldn't match the type
- [HappyParser.Token'] against PrelBase.Int
- Expected: HappyParser.HappyReduction
- Inferred: PrelBase.Int -> HappyParser.Token' -> HappyParser.HappyState HappyParser.Token' ([HappyParser.HappyAbsSyn] -> [AbstractSyntax.Production]) -> PrelBase.Int -> PrelBase.Int -> o{-a1yN-} -> o{-a1yO-} -> [HappyParser.Token'] -> a{-a1yP-}
- In an equation for function HappyParser.action_1:
- HappyParser.action_1 _ = HappyParser.happyFail
+f :: forall a. a->a
+f :: forall a. a->a -- OK
+f :: forall a. [a] -> [a]
+f :: forall b. b->b = e ???
-* GHC_ONLY/bugs/andy_cherry
-DataTypes.lhs: 3: Could not find valid interface file for `GenUtils'
+-------------------------------
+NB: all floats are let-binds, but some non-rec lets
+ may be unlifted (with RHS ok-for-speculation)
-Need "make depend"
-* GHC_ONLY/bugs/lex
+simplArg: [use strictness]
+ [used for non-top-lvl non-rec RHS or function arg]
+ if strict-type || demanded
+ simplStrictExpr
+ else
+ simplExpr ---> (floats,expr)
+ float all the floats if exposes constr app, return expr
-Pattern match fail in lex; must be producing empty or multi-valued result
+simpl (applied lambda) ==> simplNonRecBind
+simpl (Let (NonRec ...) ..) ==> simplNonRecBind
-Aggravated by dreadful error messages:
-+
-+ Fail: In irrefutable pattern
-+ Fail: In pattern-matching
-+ Fail: In pattern-matching
-+ Fail: In pattern-matching
-+ Fail: In pattern-matching
-+ Fail: In pattern-matching
-+ Fail: In pattern-matching
-+ Fail: In pattern-matching
-+ Fail: In pattern-matching
-+ Fail: In pattern-matching
-+ Fail: In pattern-matching
-+ Fail: In pattern-matching
-+ Fail: In pattern-matching
-+ Fail: In pattern-matching
-+ Fail: In pattern-matching
-+ Fail: In pattern-matchingtoo many nested calls to `error'
+simpl (Let (Rec ...) ..) ==> simplRecBind
+simplRecBind:
+ simplify binders (but not its IdInfo)
+ simplify the pairs one at a time
+ using simplRecPair
-* GHC_ONLY/bugs/jtod_circint
+simplNonRecBind: [was simplBeta]
+ [used for non-top-lvl non-rec bindings]
+ - check for PreInlineUnconditionally
+ - simplify binder, including its IdInfo
+ - simplArg
+ - if strict-type
+ addCaseBind [which makes a let if ok-for-spec]
+ else
+ completeLazyBind
-Main.hs: 12: No instance for: Signal.Signal (Signal.Stream Bit.Bit)
- Main.hs: 12: at a use of an overloaded identifier: `Signal.one'
+simplLazyBind: [binder already simplified, but not its IdInfo]
+ [used for both rec and top-lvl non-rec]
+ [must not be strict/unboxed; case not allowed]
+ - check for PreInlineUnconditionally
+ - substituteIdInfo and add result to in-scope
+ [so that rules are available in rec rhs]
+ - simplExpr --> (floats,expr)
+ - float: lifted floats only
+ if exposes constructor or pap (even if non-triv args)
+ or if top level
+ - completeLazyBind
+
-instance-decl slurping is WRONG
+completeLazyBind: [given a simplified RHS]
+ [used for both rec and non-rec bindings, top level and not]
+ - try discarding dead
+ - try PostInlineUnconditionally
+ - let-bind coerce arg and repeat
+ - try rhs tylam (float)
+ - try eta expand (float) [not if any float is unlifted && (non-spec || top_lvl || rec)]
+ - let-bind constructor args [not if any float is ..as above..]
-* GHC_ONLY/arith005
+ - add unfolding [this is the only place we add an unfolding]
+ add arity
-ceiling doesn't work properly
---- 1,3 ----
-+ [1, 1, 2, 3, 4, 5, 0, -2, -3, -4, 1000013, 124, 101, 103, 1, 0, 17001, 0, 1, 4]
-+ [1, 1, 2, 3, 4, 5, 0, -2, -3, -4, 1000013, 124, 101, 103, 1, 0, 17001, 0, 1, 4]
- [0, 0, 2, 3, 4, 5, -1, -2, -3, -4, 1000012, 124, 101, 103, 1, 0, 17000, 0, 1, 4]
-***************
-*** 2,5 ****
- [0, 0, 2, 3, 4, 5, -1, -2, -3, -4, 1000012, 124, 101, 103, 1, 0, 17000, 0, 1, 4]
-- [0, 0, 2, 3, 4, 5, -1, -2, -3, -4, 1000012, 124, 101, 103, 1, 0, 17000, 0, 1, 4]
-- [0, 0, 2, 3, 4, 5, -1, -2, -3, -4, 1000012, 124, 101, 103, 1, 0, 17000, 0, 1, 4]
- [0, 0, 1, 2, 3, 4, -1, -3, -4, -5, 1000012, 123, 100, 102, 0, -1, 17000, -1, 0, 3]
---- 4,5 ----
+Right hand sides and arguments
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+In many ways we want to treat
+ (a) the right hand side of a let(rec), and
+ (b) a function argument
+in the same way. But not always! In particular, we would
+like to leave these arguments exactly as they are, so they
+will match a RULE more easily.
+
+ f (g x, h x)
+ g (+ x)
-* GHC_ONLY/bugs/lennart_array
+It's harder to make the rule match if we ANF-ise the constructor,
+or eta-expand the PAP:
-Wrong array semantics (but who cares?)
+ f (let { a = g x; b = h x } in (a,b))
+ g (\y. + x y)
-* GHC_ONLY/bugs/life_space_leak
+On the other hand if we see the let-defns
--n *** sum I got:
-0 0
--n *** sum I expected:
-02845 1350
+ p = (g x, h x)
+ q = + x
+
+then we *do* want to ANF-ise and eta-expand, so that p and q
+can be safely inlined.
+
+Even floating lets out is a bit dubious. For let RHS's we float lets
+out if that exposes a value, so that the value can be inlined more vigorously.
+For example
+
+ r = let x = e in (x,x)
+
+Here, if we float the let out we'll expose a nice constructor. We did experiments
+that showed this to be a generally good thing. But it was a bad thing to float
+lets out unconditionally, because that meant they got allocated more often.
+
+For function arguments, there's less reason to expose a constructor (it won't
+get inlined). Just possibly it might make a rule match, but I'm pretty skeptical.
+So for the moment we don't float lets out of function arguments either.
+
+
+Eta expansion
+~~~~~~~~~~~~~~
+For eta expansion, we want to catch things like
+
+ case e of (a,b) -> \x -> case a of (p,q) -> \y -> r
+
+If the \x was on the RHS of a let, we'd eta expand to bring the two
+lambdas together. And in general that's a good thing to do. Perhaps
+we should eta expand wherever we find a (value) lambda? Then the eta
+expansion at a let RHS can concentrate solely on the PAP case.
projects / ghc-hetmet.git / blobdiff