Re: avoiding unwanted partial responses

On Tue, Aug 26, 2014 at 07:52:37AM +0200, Carlo Capelli wrote:
> member(E,L) with unbound L it's the cause of those always longer lists. I
> think you can simplify the code, maybe in this way
> 
> /* added for test
> 
> mount('Dev', 'Path', 'W1').
> 
> mount('Dev', 'Path', 'W2').
> 
> */
> 
> 
> dup_mounts(Results):-
> 
>   findall(m(Dev, Path), (mount(Dev, Path, W1), mount(Dev, Path, W2), W1 \=
> W2), DupMounts0),
> 
>   list_to_ord_set(DupMounts0, DupMounts),
> 
>   maplist(dup_mount, DupMounts, Results).
> 
> 
> dup_mount(m(Dev, Path), Whens) :-
> 
>   setof(When, mount(Dev, Path, When), Whens).

Thanks; that worked great.

Apparently there were only 2 duplicates, and so if I had just stopped
with the first response from the old code I would have had the answer.

Ross
> 
> 
> 
> 
> 
> 
> 2014-08-26 4:54 GMT+02:00 Ross Boylan <[email protected]>:
> 
> > The code below attempts to generate a single set of Results with all its
> > elements.
> > When run it instead gives multiple answers with a partially unresolved
> > set; it seems to simply
> > increase the number of unknown elements each time.
> >
> > First, I'd like to understand why it's behaving as it does.
> > Second, more importantly, what is the idiomatic way to achieve the result
> > I want?
> >
> > % -Results has all occurrence of multiple mounts with same device and path
> > % Results is a list, each element of which is m(Dev, Path)-Whens
> > % where Whens is a sorted list of mount times.
> > dup_mounts(Results):-
> >         findall(m(Dev, Path), (mount(Dev, Path, W1), mount(Dev, Path, W2),
> > W1 \= W2), DupMounts0),
> >         list_to_ord_set(DupMounts0, DupMounts),
> >         dm1(DupMounts, Results).
> >
> > dm1([], _Results).
> > dm1([m(Dev, Path) | Rest], Results) :-
> >             setof(When, mount(Dev, Path, When), Whens),
> >             member(m(Dev, Path)-Whens, Results),
> >             dm1(Rest, Results).
> >
> > When run produces
> > ?- dup_mounts(R).
> > R = [m(pdd1, oscorn1~/ms/win32)-[2006-2-6 #50, 2007-12-1 #120], m(pdd5,
> > oscorn1~/ms/ntfs)-[2006-2-6 #40, 2007-12-1 #110]|_G703] ;
> > R = [m(pdd1, oscorn1~/ms/win32)-[2006-2-6 #50, 2007-12-1 #120], _G702,
> > m(pdd5, oscorn1~/ms/ntfs)-[2006-2-6 #40, 2007-12-1 #110]|_G706] ;
> > R = [m(pdd1, oscorn1~/ms/win32)-[2006-2-6 #50, 2007-12-1 #120], _G702,
> > _G705, m(pdd5, oscorn1~/ms/ntfs)-[2006-2-6 #40, 2007-12-1 #110]|_G709] ;
> > R = [m(pdd1, oscorn1~/ms/win32)-[2006-2-6 #50, 2007-12-1 #120], _G702,
> > _G705, _G708, m(pdd5, oscorn1~/ms/ntfs)-[2006-2-6 #40, 2007-12-1
> > #110]|_G712] ;
> > R = [m(pdd1, oscorn1~/ms/win32)-[2006-2-6 #50, 2007-12-1 #120], _G702,
> > _G705, _G708, _G711, m(pdd5, oscorn1~/ms/ntfs)-[2006-2-6 #40, 2007-12-1
> > #110]|_G715] ;
> > R = [m(pdd1, oscorn1~/ms/win32)-[2006-2-6 #50, 2007-12-1 #120], _G702,
> > _G705, _G708, _G711, _G714, m(pdd5, oscorn1~/ms/ntfs)-[2006-2-6
> > #40|...]|_G718] ;
> > R = [m(pdd1, oscorn1~/ms/win32)-[2006-2-6 #50, 2007-12-1 #120], _G702,
> > _G705, _G708, _G711, _G714, _G717, m(..., ...)-[...|...]|_G721] ;
> > R = [m(pdd1, oscorn1~/ms/win32)-[2006-2-6 #50, 2007-12-1 #120], _G702,
> > _G705, _G708, _G711, _G714, _G717, _G720, ... - ...|...] .
> > % I stopped it here because it wasn't getting more helpful.
> >
> > I've accomplished similar things by having a list of partial results that
> > gets appended to at each iteration, but I was hoping
> > for something more elegant.
> >
> > BTW, printing out each solution like
> > m(pdd1, oscorn1~/ms/win32)-[2006-2-6 #50, 2007-12-1 #120]
> > is in some ways better, but I'm also interested in getting the results
> > into a structure I could work with further.
> >
> > Ross Boylan
> >
> > P.S.  Thanks, Richard, for answering my earlier question.  Should I file a
> > bug report or tag the relevant pages in the online
> > manual?
> > _______________________________________________
> > SWI-Prolog mailing list
> > [email protected]
> > https://lists.iai.uni-bonn.de/mailman/listinfo.cgi/swi-prolog
> >