[m-dev.] for review: rewrite of termination analysis (part 2)

[Christopher Rodd SPEIRS]

> +
> + at sp 1
> + at item --termination-single-argument-analysis @var{limit}
> + at itemx --term-single-arg @var{limit}
> +When performing termination analysis, try analyzing
> +recursion on single arguments in strongly connected
> +components of the call graph that have up to @var{limit} procedures.

Is it worth documenting that "setting this limit to zero disables single
argument termination analysis"?


>  
> -	% Given a bag, produce a sorted list with no duplicates 
> +	% Given a bag, produce a sorted list containing all the values in
> +	% the bag.  Each value will appear in the list once, with the
> +	% associated value giving the number of times that it appears
> +	% in the bag.
> +:- pred bag__to_assoc_list(bag(T), assoc_list(T, int)).
> +:- mode bag__to_assoc_list(in, out) is det.
> +
This comment should be clarified.  Currently `value' is used to refer to
both the values stored in the bag, and the integer describing the number
of times each value occurs.  

> +
> +list__take_upto(N, As, Bs) :-
> +	(
> +		N > 0
> +	->
> +		N1 is N - 1,
> +		(
> +			As = [A | As1],
> +			Bs = [A | Bs1],
> +			list__take_upto(N1, As1, Bs1)
> +		;
> +			As = [],
> +			Bs = []
> +		)
>  	;
>  		Bs = []
>  	).

Is there a particular reason that this predicate is not implemented as
list__take_upto(N, As, Bs) :-
	( list__take(N, As, Bs0) ->
		Bs = Bs0
	;
		Bs = As
	).

It just seems like unnecessary code duplication.

Chris