[m-rev.] for review: allow tracking of input subterms through builtin_catch

[Julien Fischer]

On Thu, 8 Dec 2005, Ian MacLarty wrote:

> For review by anyone.
>
> Estimated hours taken: 5
> Branches: main
>
> Allow input subterms to be tracked through builtin_catch.  By making

I suggest: This is done by making a special case ...


> a special case of catch_impl, which is the only predicate that calls
> builtin_catch (we have to make catch_impl the special case, because
> builtin_catch doesn't generate any events, so has no node in the EDT).
>
> browser/declarative_tree.m:
> 	Break trace_dependency into two smaller predicates.
> 	If tracking an input to builtin_catch then return what we know
> 	the origin will be instead of invoking the usual subterm dependency
> 	tracking algorithm.
>
> library/exception.m:
> 	Note that the debugger assumes builtin_catch is only called from
> 	catch_impl.
>
> tests/debugger/declarative/Mmakefile:
> tests/debugger/declarative/track_through_catch.exp:
> tests/debugger/declarative/track_through_catch.inp:
> tests/debugger/declarative/track_through_catch.m:
> 	Test the change.
>
> Index: browser/declarative_tree.m
> ===================================================================
> RCS file: /home/mercury1/repository/mercury/browser/declarative_tree.m,v
> retrieving revision 1.35
> diff -u -r1.35 declarative_tree.m
> --- browser/declarative_tree.m	7 Dec 2005 16:06:59 -0000	1.35
> +++ browser/declarative_tree.m	7 Dec 2005 21:12:25 -0000

...


> @@ -893,6 +868,67 @@
>  		Mode = subterm_out
>  	).
>
> +:- pred trace_dependency_in_proc_rep(S::in, term_path::in,
> +	start_loc(R)::in, int::in, int::in, R::in, maybe(goal_path)::in,
> +	proc_rep::in, subterm_origin(edt_node(R))::out)
> +	is det <= annotated_trace(S, R).
> +
> +trace_dependency_in_proc_rep(Store, TermPath, StartLoc, ArgNum,
> +	TotalArgs, NodeId, StartPath, ProcRep, Origin) :-
> +	det_trace_node_from_id(Store, NodeId, Node),
> +	materialize_contour(Store, NodeId, Node, [], Contour0),
> +	(
> +		StartLoc = parent_goal(CallId, CallNode),
> +		Contour = list.append(Contour0,
> +			[CallId - CallNode])
> +	;
> +		StartLoc = cur_goal,
> +		Contour = Contour0
> +	),
> +	ProcRep = proc_rep(HeadVars, GoalRep),
> +	is_traced_grade(AllTraced),
> +	MaybePrims = make_primitive_list(Store,
> +		[goal_and_path(GoalRep, [])],
> +		Contour, StartPath, ArgNum, TotalArgs,
> +		HeadVars, AllTraced, []),
> +	(
> +		MaybePrims = yes(primitive_list_and_var(
> +			Primitives, Var, MaybeClosure)),
> +		%
> +		% If the subterm is in a closure argument then the argument
> +		% number of the closure argument is prefixed to the term path,
> +		% since the closure is itself a term.  This is done here
> +		% because at the time of the closure call it is not easy to
> +		% decide if the call is higher order or not, without repeating
> +		% all the work done in make_primitive_list.
> +		%
> +		(
> +			MaybeClosure = yes,
> +			AdjustedTermPath = [ArgNum | TermPath]
> +		;
> +			MaybeClosure = no,
> +			AdjustedTermPath = TermPath
> +		),
> +		traverse_primitives(Primitives, Var,
> +			AdjustedTermPath, Store, ProcRep,
> +			Origin)
> +	;
> +		MaybePrims = no,
> +		Origin = not_found
> +	).
> +
> +	% proc_rep_is_catch_impl(ProcRep) is true if ProcRep is a
> +	% representation of exception.catch_impl (the conserve

s/conserve/converse/

> +	% is true assuming exception.builtin_catch is only called from
> +	% exception.catch_impl).
> +	%
> +:- pred proc_rep_is_catch_impl(proc_rep::in) is semidet.
> +
> +proc_rep_is_catch_impl(ProcRep) :-
> +	ProcRep = proc_rep([A, B, C, D], atomic_goal_rep(_, "exception.m", _,
> +		[D], plain_call_rep("exception", "builtin_catch",
> +			[A, B, C, D]))).
> +
>  :- pred find_chain_start(S::in, R::in, arg_pos::in, term_path::in,
>  	dependency_chain_start(R)::out) is det <= annotated_trace(S, R).
>

The looks fine otherwise.

Cheers,
Julien.
--------------------------------------------------------------------------
mercury-reviews mailing list
post:  mercury-reviews at cs.mu.oz.au
administrative address: owner-mercury-reviews at cs.mu.oz.au
unsubscribe: Address: mercury-reviews-request at cs.mu.oz.au Message: unsubscribe
subscribe:   Address: mercury-reviews-request at cs.mu.oz.au Message: subscribe
--------------------------------------------------------------------------