Hi,
Here is a simple test that shows different behavior in Universe vs
jBase. I would like to know what is the behavior on other systems
PICK, Cache, Revelation etc.
Which behavior do you think makes more sense ?
A = 'ABABA'
B = 'ABA'
CRT A,B
FOR I = 1 TO 5
POS = INDEX(A,B,I)
WHILE POS
CRT I:' POS=':POS
NEXT I
CRT 'CNT=':COUNT(A,B)

Universe:
ABABA ABA
1 POS=1
CNT=1

jBase:
ABABA ABA
1 POS=1
2 POS=3
CNT=2

<lucian_pata (AT) yahoo (DOT) com> wrote

The first case, as it counts only distinct occurrences of 'ABA'
in 'ABABA' whereas the second case also counts the overlapping
(non-distinct) occurrence.

My uniData gives the second case; perhaps due to a UDT setting?

--
frosty

Hi all,

Unidata is documented as saying the in BasicTypes P and M, it allows
overlapping substrings. The D3 documentation doesn't mention it but I
assume that Unidata behaves as it does in these modes for
compatibility.

UniVerse has a compiler option to turn on/off overlapping substrings.

QM does not allow overlapping substrings. If someone needed it, I'm
sure we could make it an option.

Historically, Prime Information and PI/open dod no tallow overlapping
substrings.

Don't know about the rest.


Martin Phillips, Ladybridge Systems

Reality counts the overlapping substrings; the output is the same as
the jBase example.

Our ancient version of Reality doesn't support the WHILE clause in a
FOR/NEXT loop, so I had to modify the program slightly.


Rob Allen
ADP