[mercury-users] inst of list in non-empty branch

[Julien Fischer]

On Sat, 31 Mar 2007, Ian MacLarty wrote:

> On Fri, Mar 30, 2007 at 12:36:17PM +1000, Ian MacLarty wrote:
>> On Thu, Mar 29, 2007 at 02:35:10PM -0400, doug.auclair at logicaltypes.com wrote:
>>> Dear Ralph (and all), you and I wrote:
>>>
>>>>> print_kids(Element, Kids@[_|_], !IO) :-
>>>>>          format("The child element%s of ", [s(singular_plural(Kids))], !IO),
>>>>>          print(Element, !IO),
>>>>>          is_are(Kids, !IO),
>>>>>          print(Kids, !IO),
>>>>>          nl(!IO).
>>>
>>>> The mode for format/4 is pred(in, in, di, uo), so while the compiler
>>>> knows that Kids has an inst matching non_empty_list before the call to
>>>> format, it only knows it has inst ground afterwards.
>>>
>>> Okay.  I think, however, that the mode for Kids is being set to just 'in'
>>> anyway, because when I reduce the clause to:
>>>
>>> print_kids(_, Kids@[_|_], !_IO) :-
>>>         TheS = singular_plural(Kids).
>>>
>>
>> I think that:
>>
>> print_kids(_, Kids, !IO) :-
>>     Kids = [_ | _],
>>     TheS = singular_plural(Kids).
>>
>> will do the trick.
>>
>
> The following should also work:
>
> print_kinds(_, [_|_] @ Kids, !_IO) :-
> 	...
>
> (i.e. swap around the arguments of the unification expression).
>
> It seems that print_kids(_, Kids @ [_ | _], ...) is transformed into:
>
> print_kids(_, H1, ...) :-
> 	H1 = Kids,
> 	H1 = [_ | _],
> 	...
>
> while print_kids(_, [_ | _] @ Kids, ...) is transformed into:
>
> print_kids(_, H1, ...) :-
> 	H1 = [_ | _],
> 	H1 = Kids,
> 	...
>
> in the first case Kids ends up being ground, while in the second case it gets
> the inst non_empty_list.  I'm not sure why this is the case though.

That's a bit annoying.

Julien.
--------------------------------------------------------------------------
mercury-users mailing list
Post messages to:       mercury-users at csse.unimelb.edu.au
Administrative Queries: owner-mercury-users at csse.unimelb.edu.au
Subscriptions:          mercury-users-request at csse.unimelb.edu.au
--------------------------------------------------------------------------