Watch out for this discrepancy between Universe and D3
Universe does not insert null using the -1 in a replace.

Given the following program

AM=CHAR(254)
VM=CHAR(253)
A=1:VM:2:VM:3:VM:VM:5:VM:6:VM:VM:VM:9
A:=AM:'A':VM:'B':VM:'C':VM:VM:'E':VM:'F':VM:VM:VM: 'I'
A:=AM:'A1':VM:'B2':VM:'C3':VM:VM:'E5':VM:'F6':VM:V M:VM:'I9'
X=''
C=DCOUNT(A<1>,VM)
FOR J=1 TO C
FOR I=1 TO 3
X<I,-1>=A<I,J>
NEXT I
NEXT J
FOR J=1 TO C
CRT J:
FOR I=1 TO 3
CRT ' ':X<I,J>:
NEXT I
CRT
NEXT J

On D3 the output is
1 1 A A1
2 2 B B2
3 3 C C3
4
5 5 E E5
6 6 F F6
7
8
9 9 I I9

But on Universe it is
1 1 A A1
2 2 B B2
3 3 C C3
4 5 E E5
5 6 F F6
6 9 I I9
7
8
9

Ouch! That one caused me some bother.

Jeremy Thomson

This is one of those "bugs" that's been around since < R83 that we all
laughed about when they said, "that's not a bug, that's a feature" or when
we said we didn't want to change it because it would "break" otherwise
working code.

Logically does this make sense: Add nothing to the end of nothing to get
something this neither nothing (null) nor anything at all (it's a lone
attribute mark).

Something in the back of your head says, "Nothing plus nothing equals
nothing". But somehow the working public has decided that adding a null to
the end of a dynamic array somehow should make the array "different".

THIS is why there are different flavors.

Mark Brown



<panzerboy (AT) gmail (DOT) com> wrote

$OPTION EXTRA.DELIM solves that problem for you.

If you wish to compile your code in PICK FLAVOUR either modify the
BASIC VOC entry with the PICK.FORMAT or add the $OPTION PICK to the
code.

I have many applications for this and definitely see it as a feature.
I had no idea about the Universe behavior so this thread has been
helpful.

T

"Mark Brown" wrote:

For those of us who must program on multiple platforms, the obvious solution
is to _always_ find a way to produce a real number for the append. It will
never become an issue that way. And don't depend on any given platform to
forever "default" to allowing/disallowing any "assumed" behavior. If it's
not explicitly documented, it may come and go with any point release. And
even if it _is_ documented, I have found that it's always safer to code as
conservatively as possible, both for readability, and for bug/feature-change
avoidance.

Ed

<panzerboy (AT) gmail (DOT) com> wrote

What a bunch of topposters people are here :-)
Oh well, when in Rome...
I agree with Ed. Back when I did a lot of cross platform work I would
never use the <-1>.
Having spent 13 years in one environment I've gotten out of the habit.
Still theres these new fangled $OPTIONS compiler directives these days.
I've set up an include file "compileroptions.inc" on the D3 and
Universe machines.
The D3 machine has nothing but a comment in it.
The Universe one has
$OPTIONS EXTRA.DELIM
$OPTIONS R83.LOCATE

Now I can have the same program code on the different machines & they
do the same thing. Coo-ol.
Back in 'the days' used have a locate subroutine that had the proper
format for Altos, GA, Universe, Reality. 75% or it was commented out
lines of the locate format for the other machines. :-)

Ed Sheehan wrote:
My first top posted reply, I feel kinda... dirty.

Jeremy Thomson

<panzerboy (AT) gmail (DOT) com> wrote









































buch of irrelevant stuff....









buch of irrelevant stuff....



























Thank God and welcome to the club. Now don't you think this bottom post is a
waste of time and effort?
Chandru Murthi

murthi wrote:
[snip]
Did you mean to write "this botton posting?"

I would wholeheartedly agree with what you
wrote! =`:^°

--
frosty

Took you over a month to scroll down, did it ?

Chandru

"Tom Rauschenbach" <tomsusenet (AT) tomsdomain (DOT) org> wrote

LOL. Good one.

"murthi" <c_xyz_murthi (AT) seeing_xyz_green (DOT) net> wrote