October 2006 - Cvs-nhc98

patch applied (hat): The renamer for -hat did not handle the case that a locally defined
by Malcolm Wallace 10 Oct '06

10 Oct '06

Thu May 16 09:44:38 PDT 2002 malcolm * The renamer for -hat did not handle the case that a locally defined variable can shadow a field name. M ./src/compiler98/AuxLabelAST.hs -12 +21 M ./src/hattrans/AuxLabelAST.hs -12 +21

1 0

patch applied (hat): (==) should be a valid pattern in hat-observe.
by Malcolm Wallace 10 Oct '06

10 Oct '06

Thu May 16 04:27:39 PDT 2002 malcolm * (==) should be a valid pattern in hat-observe. M ./src/hattools/Pattern.hs -1 +2

1 0

patch applied (hat): Replaced definition of wrapForward by def. of wrapUForward.
by Malcolm Wallace 10 Oct '06

10 Oct '06

Fri May 10 12:11:48 PDT 2002 olaf * Replaced definition of wrapForward by def. of wrapUForward. This combinator is always used in an untraced context. Hence it is unnecessary to have two variants. Actually expected slightly smaller traces and less performance with nhc98, but there seems to be no difference. M ./src/hatlib/Hat.hs -20 +14

1 0

patch applied (hat): Changed combinators for trusting in -hat.
by Malcolm Wallace 10 Oct '06

10 Oct '06

Fri May 10 04:19:36 PDT 2002 olaf * Changed combinators for trusting in -hat. The problem with the previous implementation was that the result of the reduction of a trusted function often contained many Hiddens, making the expression rather useless. E.g.: foldr (:) [] "Hi" = ({_} ({_} [])). The cause was that wrapUForward only works if the trace is demanded before the wrapped expression is demanded. However, the uapn combinators required the function value first and sometimes later required the trace of the function. Have changed uapn combinators to never demand the trace of the function. As a result, the result of the reduction of a trusted function is always shown as Hidden, if it is of functional type. However, if it is of an algebraic type, it is shown correctly. To achieve this, the cnn combinators for partially applied constructors had to be modified to record the complete constructor application in the trace, if called from untraced code. Furthermore made the various application combinators less strict in Fun, to avoid potential blackholes. Thought this would avoid blackhole in Clausify program, but unfortunately it does not. Slightly improves performance though. M ./src/hatlib/Hat.hs -199 +319

1 0

patch applied (hat): In Hat combinator library reintroduce stack for entered redexes and use a buffer only for deferred updating of reduced redexes.
by Malcolm Wallace 10 Oct '06

10 Oct '06

Mon May 6 04:31:29 PDT 2002 olaf * In Hat combinator library reintroduce stack for entered redexes and use a buffer only for deferred updating of reduced redexes. The main advantage is that we obtain a good entry point for hat-trail into the trace if the computation was interrupted or aborted with an error that was only caught as an exception in Haskell or on the C-side. The entry point is the last redex that was entered but not yet reduced. The reference of a reduced redex is always compared with the reference on top of the stack to recognise the situation where a Haskell exception renders part of the stack garbage. Then the stack is popped until an entry with the same reference is found. The popped entries (except for the last) are written as Entered into the trace. However, it seems there are never any such entries anyway, because IO-actions always reduce successfully to Abstract "IO". Need good test case. Interruptions often happen at an entered Forward. These do not however have a parent in the trace. Hence the combinator library at least keeps a parent for every Forward on the stack, so that the entry point can be that parent. (Actually unevaluated and entered Forwards should have a parent; but that needs change of trace file format). Stack plus buffer also seems to be slightly more efficient than the previous buffering method. M ./src/hatlib/Hat.hs -3 +4 M ./src/hatlib/hat-c.c -59 +204 M ./src/hatlib/hat-c.h -4 +7 M ./src/hattools/Makefile -4 +4

1 0

patch applied (hat): Quote the function name passed to hat-observe from hat-trail.
by Malcolm Wallace 10 Oct '06

10 Oct '06

Fri May 3 09:28:21 PDT 2002 malcolm * Quote the function name passed to hat-observe from hat-trail. (To avoid the shell interpreting <= as file redirection etc.!) M ./src/hattools/HatTrail.hs -1 +1

1 0

patch applied (hat): Change (f . g) = \x-> f (g x) to (.) f g x = f (g x)
by Malcolm Wallace 10 Oct '06

10 Oct '06

Fri May 3 08:22:36 PDT 2002 malcolm * Change (f . g) = \x-> f (g x) to (.) f g x = f (g x) for nicer traces. (The literal lambda looks ugly.) M ./src/hatlib/PreludeBasic.hs -1 +1

1 0

patch applied (hat): Now obsolete.
by Malcolm Wallace 10 Oct '06

10 Oct '06

Fri May 3 04:20:24 PDT 2002 malcolm * Now obsolete. R ./src/hattools/HatExpressionTree.hs R ./src/hattools/HatTrace.gc R ./src/hattools/PrettyLib.hs

1 0

patch applied (hat): Comment out the over-large display squashing code. Reduce the maximum
by Malcolm Wallace 10 Oct '06

10 Oct '06

Fri May 3 04:11:00 PDT 2002 malcolm * Comment out the over-large display squashing code. Reduce the maximum size of the output pane to 1/4 of full screen, rather than 1/3. M ./src/hattools/HatTrail.hs -3 +4

1 0

patch applied (hat): Fix incorrect pretty-printing indent in hat-observe: the highlighting
by Malcolm Wallace 10 Oct '06

10 Oct '06

Fri May 3 04:10:01 PDT 2002 malcolm * Fix incorrect pretty-printing indent in hat-observe: the highlighting of the eqn number should be done *after* determining its length. M ./src/hattools/HatObserve.hs -2 +2 M ./src/hattools/SExp.hs -2 +3

1 0