Re: Some thoughts on dicts in SWIPL 7

On 02/18/2014 11:58 AM, Carlo Capelli wrote:
> 
> 
> 
> 2014-02-18 11:48 GMT+01:00 Jan Wielemaker <[email protected]
> <mailto:[email protected]>>:
> 
>     On 02/18/2014 11:27 AM, Carlo Capelli wrote:
>     >
>     > I forgot: a pattern that could be handy could be a member_dict/2, to
>     > avoid this kind of code
>     >
>     > (member(F, T.foreignKeys), F >:< _{pkCol:PkCol, fkTable:FkTable,
>     > fkCol:FkCol})
> 
>     I think I'm missing something:
> 
>     16 ?- A is _{a:1, b:2}.Key.
>     A = 1,
>     Key = a ;
>     A = 2,
>     Key = b.
> 
>     What am I missing?
> 
> 
> What I meant: porting older code, that was like
> 
> table(T, ..., ForeignKeys),
> member(foreignKey(PkCol,FkTable,FkCol), ForeignKeys)
> 
> I initially wrote
> 
> member(_{pkCol:PkCol, fkTable:FkTable,fkCol:FkCol}, T.foreignKeys),
> 
> and that worked. Then I realized that if I add a field to foreignKeys,
> it will fail...

Yes.  You mean T.foreignKeys is a list of dicts and you want to do a
member and then only get dicts that have at least a certain set of
keys?  That would apply for any extraction.  The psi-term ideas are
the way to go here, otherwise we'll end up with a (sub-)dict version
for all access predicates ...

	Cheers --- Jan