Hi

I remember seeing a tip somewhere years ago on a website for a really
simple synonym which found the last multivalue in a multivalued field,
it did it straight off without invoking any subroutine. Can someone
advise me of the syntax required please.

regards

John Marshall

johnmarshall (AT) xtra (DOT) co.nz wrote:
Google's good. I found this by googling cdp for "last mv"...

http://tinyurl.com/jcszg

Mike

thanks for the link. I'd searched Google on the web for the answer but
had not thought of CDP - duh!

Anyway, I tried what was recommended and it works great.

thanks

John
Mike Preece wrote:

johnmarshall (AT) xtra (DOT) co.nz wrote:
AS slimy hacks go, yeah. I find it difficult to actually approve of it,
however. Fortunately I've never had to use it myself, because I'd
probably have to scrub like Lady Macbeth afterwards. ;^)

Luke

Tut, Mike, you're not following established cdp standards, which require you
to say:

'I found this by googling cdp for "last mv"...'

but NOT give the url. Don't make it too easy for John!

Chandru Murthi
ref : news:mmm4d2pnk3er60oscd1red2norlarafpea (AT) 4ax (DOT) com

"Mike Preece" <michael (AT) preece (DOT) net> wrote

"Mike Preece" wrote:

(gotta love tinyurl...)

RE: Syntax:
f;3r
or f;3;p;s;_
or F;{atb};{atb};S;_

This is one of those things that we know works but for my entire
career I've avoided asking exactly why it works. I really want to
understand what's happening here. Though as Luke suggests I would
itch terribly if I ever used it because it's not documented and
subject to break in any release on any platform.

Can some guru please provide some insight into exactly what makes this
syntax tick? I think Henry could write a couple pages on this. Jon,
I don't suppose you could prod him to make a guest appearance?
Chandru, was this something of yours or Tim's?

Thanks!
T

I'm currently on a Linux/Universe 10.1.4 (pick format) system.
f;3;p;s;_ on this release returns all of the multivalues.
f;3;s;p;_ returns a sum of the numeric data elements (ignoring non-numerics)
The results are the same for "3r".
Confused ... what versions/systems does this work on?

"Tony Gravagno" <g6q3x9lu53001 (AT) sneakemail (DOT) com.invalid> wrote

Tom Phillips wrote:
D3 and its predecessors, AP, R83 and all their various avatars
(Ultimate, mvBase, mvEnterprise etc).

Luke

"Tony Gravagno" <g6q3x9lu53001 (AT) sneakemail (DOT) com.invalid> wrote in message
Mr T,

I may have this backwards; it's been years since I've seen the source code.

When you're attempting to match up values in attributes that might not have
the same number of values, there are several things you need to be able to
do:

(R) repeat one value against all the values of another attribute. For
example, to concatenate a single valued "prefix" to multivalues to make a
remote key, something like this: f;1r;2;: repeats the single value in att1
with each value in attribute 2

(S) Sum all the values, run thru the values of an attribue and accumulate a
single value.

(P) Push a value from the stack onto the stack

(_) Swap two values on the stack.

So f;3r probably pushes the first value onto the stack and repeats it as
required and since it's only required once, that's the one you get.
f;3;p;s;_ pushes the value onto the stack(3), pushes the top value again
(p), performs a sum which pushes a "total" onto the stack (s) and then swaps
the two top values leaving that (P)ushed value now on top of the stack.

The stack is like the little plastic coin holders they have for people who
do a lot of tolls. You push coins into the top and you pop them out the
top. Each "command" code (eg. +, -, :, etc) pops as many items off the
stack as required and pushes the answer back. So f;3;4;+ pushes attr 3,
pushes attr 4; then the + pops off two values, adds them togethe and pushes
the answer back.

The problem with these correlatives is that they leave unwanted/unused items
on the stack, like the f;3;p;s;_ will leave the sum on the stack, probably 0
if the attribute is non-numeric. With systems that have limited stack
space, that can be a problem.

Mark