Hi Don,
VARYING means that the variable will start with two bytes that indicate
the current length of data, followed by X number of bytes containing the
actual data in the string.
Just to make sure the concept is clear, consider this code:
D MyString s 100a varying
Under the covers, the preceding line of code will have the EXACT SAME
layout in the computer's memory as the following data structure:
D MyString ds
D len 5u 0
D data 100a
I.e., both of them are 102 bytes of memory, both of them start with a 2
byte length, and are followed by a fixed-length 100 character chunk of
memory containing the data of the string. VARYING does *NOT* mean that
the total amount of memory used is variable. A VARYING field always
uses the same amount of memory. In my example, 102 bytes.
VARYING is equivalent to SQL's VARCHAR data type, or DDS's VARLEN.
So what's wrong with your code? Well, frankly, a lot of things. First
of all, DON'T USE A POINTER THIS WAY. This is just asking for trouble.
You pass a pointer by reference, then force it to always point to a
1A VARYING. What's the point? Are you allocating the memory for this
pointer in the procedure? That's the only reason I can come up with why
you'd want to do this.... but it doesn't appear that you do (if you do,
you omitted that portion of the code.)
Instead, you pass a pointer by reference (which means your procedure can
change the memory it points to) and then you do nothing with it but
treat it as a 1A VARYING field. Just pass the 1A VARYING to begin with
if that's what you really want.
But the most important thing is this: When you point the 1A VARYING at
a spot in memory, you have to make DARNED sure that the spot in memory
contains a 1A VARYING variable. Remember, 1A VARYING occupies THREE
bytes of memory (not one) and the first two bytes contain the current
length of the data. Since your variable is 1A, that means the first teo
bytes MUST contain hex x'00000000' or x'00000001'. (Length of zero or
one.) If it contains anything else you'll have a big problem because
your field can ONLY store a length of one or zero, since it's a 1A field.
You state in your message that you sometimes get "Length of variable is
out of range". That's the reason. Because those first two bytes MUST
have hex values of zero or one. If they have anything else, you're
going to have problems.
I'd recommend an alternative, but frankly I have no clue of what you're
trying to accomplish here. This code doesn't make any sense that I can
fathom... it makes no sense whatsoever.
I've read your other messages in this thread (as of 2:30pm CST) and I
still can't make heads or tails of what you're trying to do.
Don Wereschuk wrote:
P FillArrayAll B
D FillArrayAll PI
D FieldToCheck *
D CheckValue 50A
D ArrayValue 50A
D*
DCheckField S 1A based(FieldToCheck) VARYING
[snip]
midrange.com
