REDIM vs DIM with Passed Arguments

[Pat Bruening]

"in a procedure or function, you can use REDIM to redimension
an array that was passed as a parameter."

I have found that this is true only if the original array was
first REDIMed rather than simply DIMed. If DIM is used, and I
try to REDIM a passed array, I get an error message "duplicate
array name" (or similar). However, if I REDIM the array, I can
then REDIM it within a procedure.

Anyone know why this is so?

Thanks,
Pat


------------------

[Lance Edmonds]

Pat, I suspect it has to be something particular to do with your code... can you post a small example that demonstrates your problem?

(I often redim arrays that are passed as a parameter)

Thanks!


------------------
Lance
PowerBASIC Support
mailto:[email protected][email protected]</A>

[Pat Bruening]

Lance,

Thanks for a quick response. Here is the code you requested:

FC$="Allen, Donald,Robert,Charles,"
redim c$(1)
call control(C$(),N,FC$)

sub control(c$(),N,FC$)
N=TALLY(FC$,",")
redim C$(N)
exit sub

------------------

[Lance Edmonds]

That code works fine with PB/DOS 3.5... what compiler are you using?

PB2.x and FirstBasic do not support REDIM of passed arrays.

------------------
Lance
PowerBASIC Support
mailto:[email protected][email protected]</A>

[Lance Edmonds]

Duh! I just reread your message... The problem is simple: your array is STATIC and cannot be redimmed, so the compiler flags the REDIM as an error.

The solution is to explicitly set the array to be dynamic, or set the initial boundaries using a variable (not an equate), or use a REDIM statement instead. This is a documented behavior.

$STATIC           ' the default array type will be STATIC
DIM DYNAMIC a$(1) ' dynamic as explicitly declared as such
DIM b$(1)         ' static as boundaries are known at compile-time, and $STATIC is in effect
DIM c$(%MaxSize)  ' static (as above)
DIM d$(x%)        ' dynamic as x% not resolvable at compile-time
DIM STATIC e$(n%) ' not valid - array is declared as static, but boundaries not resolvable at compile-time.
$DYNAMIC          ' the default array type will be DYNAMIC
DIM e$(1)         ' dynamic as $dynamic in effect
DIM f$(%maxsize)  ' dynamic (as above)
DIM g$(x%)        ' dynamic
DIM STATIC h$(1)  ' static as explicitly declared as such
DIM STATIC i$(n%) ' not valid - array is declared as static, but boundaries not resolvable at compile-time.

Also see $HUGE, etc.

I hope this helps...!


------------------
Lance
PowerBASIC Support
mailto:[email protected][email protected]</A>

[Mel Bishop]

I thought Pat was on to something so I wouldn't have to upgrade
from my trusty PB2.1f.

I wrote this small test program to check his posting with
absolutely no luck. I tried all possible combinations and it
still errors (599) out on the REDIM

$lib all off
color 14,1
cls

i = 4
dim a$(i)
i = i + 1
redim a$(i)

a$(1) = "The quick "
a$(2) = "brown fox "
a$(3) = "jumped over "
a$(4) = "the lazy dog "
a$(5) = "1234567890 times."

locate 1,1
for x = 1 to 5
print;a$(x);
next x
call test.data(a$(),i)
end

sub test.data(a$(),i)
' but it will work with: (adjusting parameter line)
' sub test.data(i)
' shared a$()

i = i + 1
redim a$(i)
a$(1) = "Now is the "
a$(2) = "time for all "
a$(3) = "good men to "
a$(4) = "come to the aid "
a$(5) = "of their country. "
a$(6) = "This is a padding line"

locate 3,1
for x = 1 to i
print;a$(x);
next x
end sub


------------------

[Lance Edmonds]

Sorry Mel. As I noted above, PB2.x and FirstBasic do not support REDIM of passed arrays.


------------------
Lance
PowerBASIC Support
mailto:[email protected][email protected]</A>

[Pat Bruening]

Lance and All,
Thanks. I did not consider the STATIC nature of the matrix.
Lance,
I'm using PB3.2 compiler. I just tried DIM DYNAMIC C$(1) and
it works fine.
Again, thanks for the help.
Pat


------------------

[Pat Bruening]

Lance,
The DYNAMIC DIM works with PB3.2. You mentioned that the first
way I DIMed the matrix made it STATIC. But actually, DIM doesn't
work if the matrix is made STATIC by use of a constant, as in:
DIM C$(1).
But if I use REDIM C$(1), the sub works. Does this mean that
REDIM always makes the matrix REDIMed DYNAMIC, even tho a constant
value is used (in this case, 1)? If this is so, what happens to
the string space allocated by the compiler at compile time for
the STATIC matrix (arrray)? Or is the compiler smart enough to
interpret REDIM C$(1) as a DYNAMIC array rather than as STATIC ?
I certainly would like to know what the DIM and REDIM is doing
behind the scenes.
Thanks again.
Pat


------------------


[This message has been edited by Pat Bruening (edited January 15, 2001).]

[Lance Edmonds]

Since REDIM can only be used on non-STATIC arrays, the compiler produces code for a DYNAMIC array instead.

At runtime, an array cannot be erased or redimmed as another type of array.

Therefore, you can't use the REDIM STATIC keyword combination, but you can use REDIM DYNAMIC, etc, provided the array type and data type are consistent through-out the scope of the array.

I hope this helps.

------------------
Lance
PowerBASIC Support
mailto:[email protected][email protected]</A>

[Pat Bruening]

Lance,
Thanks again for your quick response. It is quite logical that
STATIC arrays cannot be REDIMed, simply because the complier at
compile time sets aside memory; whereas DYNAMIC arrays, are, as
the name suggests, DYNAMIC (changeable). I found the answer to
my original question in the User's Guide:
DECLARING static and Dynamic arrays p.103
Arrays are dynamic in the following cases:
...
If the array appears in an ERASE or REDIM statement.

Thanks again for the help.
Understanding why is better than merely knowing what. In this
case, with your help, I now understand better the differences
between STATIC and DYNAMIC arrays.

Pat

------------------