Enrico,
Ofcourse!, My design was wrong, thankyou, thats exactly what I wanted
and I feel a bit stupid for missing it!
Ta
"Enrico Parisi" <[EMAIL PROTECTED]> wrote in message
news:[EMAIL PROTECTED]
> OK, now I understand better, it is still unclear what the query is
> supposed to return, I assume the ID.
> I assume you don't need to return the parent of changed children,
> otherwise there may be different solution.
>
> I would do the following:
>
> 1) in class A,B,C,D,E add a calculated property called HasChanged
> (returns 0/1), in the compute code you can call your function/method or
> write the code there directly. Depending on the code it might be
> possible to create an abstract class with this property and inherit
> (using multiple inheritance) in class A,B,C,D.
>
> 2) use the query:
>
> select ID
> from A
> where HasChanged = 1
>
> union
>
> select ID
> from B
> where HasChanged = 1 OR LinkToA->HasChanged=1
>
> union
>
> select ID
> from C
> where HasChanged = 1 OR OR LinkToB->HasChanged=1 OR
> LinkToB->LinkToA->HasChanged=1
>
> etc..
>
> HTH
> Ciao
>
> Enrico
>
> Paul wrote:
>
> > Hi Enrico (and anyone else listening),
> >
> > Basically I need to know if anything has changed in Class A,B,C,D or
E
> > (with each being a parent of the previous) and if something has changed
in a
> > parent I need to have ALL the children associated regardless if data has
> > changed on that child, as well as if nothing has changed on the parent
but
> > just on the child, I need that child (which will ofcourse give me the
parent
> > regardless).
> >
> > I did concider creating cursors for each and then walking down each
with
> > a cursor, this would work but I would prefer to have a query I could
call
> > instead.
> >
> > Does that make sense!?!?!?
> >
> > e.g.
> >
> > ^A(1)="AField1,AField2,AField3"
> > ^B(1,1)="BField1,BField2,BField3"
> >
> > In mumps its really easy!, But in SQL how do I implement this without
> > knowing the structure of the globals (i.e. just knowing the classes and
> > their properties)
> >
> > K ARR
> > S A="" F S A=$O(^A(A)) Q:A="" S OK=$$CHECK(^A(1)) D
> > .S B="" F S B=$O(^B(A,B)) Q:B="" I OK!($$CHECK(^B(A,B))) D SAVEIT
> >
> > CHECK() ; Code not relevent, checks to see if data changed since
previous
> > and returns 1 if has, 0 if hasn't
> > Q OK
> >
> > SAVEIT() ; Again not relevent, but some routine that saves these
nodes
> > Q
> >
> >
> >
> > "Enrico Parisi" <[EMAIL PROTECTED]> wrote in message
> > news:[EMAIL PROTECTED]
> >
> >>Hi Paul,
> >>
> >>sorry...but I still don't understand what you actually need to achive,
> >>the structure is now more clear while the goal, for me, is not!
> >>
> >>See comments inline
> >>
> >>Enrico
> >>
> >>Paul wrote:
> >>
> >>
> >>>I have 3 class each is a child of the previous, i.e.
> >>>
> >>>A has Children B
> >>>B has Children C
> >>>
> >>>I can't alter the class definitions as they are mapped to legacy
> >
> > globals,
> >
> >>>but I want to access them and produce information
> >>>based on whats in A, B and C.
> >>>
> >>>A, B and C hold data relevent to that level, and any of the data can be
> >>>changed without effecting data below.
> >>
> >>So far, so good.
> >>
> >>
> >>>Hence I have an query on A which tells me if anything has changed and
> >>>populate an array with the ID's
> >>
> >>Here I start getting confused, what is your goal? It looks you need to
> >>find out what "has changed", how do you check that? The array looks like
> >>an attempt to resolve the problem...but first, let's understand the
> >>problem :)
> >>
> >>>I then want to query B and see if anything has changed AND also return
> >>>anything which has changed on A (i.e. within the array of ID's).
> >>>
> >>>And then so forth for C (and infact D and E).
> >>>
> >>>I thought I could do the whole select with a select (x5), but the SQL
> >
> > got
> >
> >>>massive and also didn't work hence I was splitting the process down to
> >
> > do a
> >
> >>>query on each class.
> >>>
> >>>
> >>> Hope that makes sense!, ANY help much appreciated.
> >>>
> >>>
> >>>"Enrico Parisi" <[EMAIL PROTECTED]> wrote in message
> >>>news:[EMAIL PROTECTED]
> >>>
> >>>
> >>>>Hi Paul,
> >>>>
> >>>>I don't fully understand your question, what do you mean with "if its
> >>>>parent is in the array of objects" ?
> >>>>
> >>>>I cannot be precise but I think you can get advantage of Cach� implici
> >>>>joins like "Parent->ID" to get the ID of a referenced table.
> >>>>
> >>>>Maybe something like this?
> >>>>
> >>>>Select <whatever>
> >>>
> >>>>from tab3
> >>>
> >>>>where tab2->tab1->ID = ParentID
> >>>>
> >>>>In this case tab1 is parent, tab2 is child, tab3 is the SQL projected
> >>>
> >>>array.
> >>>
> >>>
> >>>>If you post a sample tables structire it would be easier to help you.
> >>>>
> >>>>Ciao
> >>>>
> >>>>Enrico
> >>>>
> >>>>Paul wrote:
> >>>>
> >>>>
> >>>>>Hello All,
> >>>>> Not being a SQL whiz I've got a "surely" simple problem, can
> >
> > someone
> >
> >>>>>help?
> >>>>>
> >>>>>
> >>>>> I have an array of objects and wish to select on a child table if
> >>>
> >>>its
> >>>
> >>>
> >>>>>parent is in the array of objects, how do I write this in SQL?
> >>>>>
> >>>>>I want in logic terms this...
> >>>>>
> >>>>>SELECT * FROM Table WHERE Field1=:VAR1 OR $D(:ARRAY(Parent->ID))
> >>>>>
> >>>>>But ofcourse $D isn't an SQL function! I can't figure how to do it in
> >>>
> >>>SQL..
> >>>
> >>>
> >>>>>Any help much appreciated.
> >>>>>
> >>>>>
> >>>>>Thankyou
> >>>>>
> >>>>>
> >>>>
> >>>
> >
> >
>
