Re: Data Structure array/Lookup questions

[Scott Klement <klemscot@xxxxxxxxxxxx>]

On Wed, 14 Apr 2004 CarollaT@xxxxxxxxxxxxxx wrote:
[SNIP]
> First of all, can I somehow DIM the data structure array based on the number
> of records in the company file?  This way, if we add companies, I won't have
> to modify and re-compile.  Also, I won't have to make a guess-timate of
> around 150, and dimension unneeded elements.
[SNIP]

With only 150 elements, I wouldn't worry much about it.  In fact, I'd go
with 300 elements just to be safe.

If for some reason you're really worried about this, you can define the
array as 30000 elements, but base it on a pointer so that no memory is
reserved.  Allocate the memory as you need it (with the %alloc() and
%realloc BIFs)...   this won't perform as well as simply declaring a 300
element array, but will be a bit more efficient on memory.

If you do use the dynamic allocation method, make sure that you never try
to access the un-allocated memory, and don't run it through an op-code
like SORTA (which will also try to access that memory.)

> Then, if the array elements contain a 3-alpha and a 11,2-packed field, can I
> use the %Lookup BIF to lookup the 3-alpha identifier for each company as I
> am summarizing?  Will the other data structure field interfere with this?

I don't think you can use %lookup() on an array of data structures or a
MODS. You may be able to use it with those overlaid array thinggies,
though.  Again, if the array is dynamic, make sure you specify a number of
elements to %lookup() so that it does not access areas that you haven't
allocated.

Are you intending to sort the array? If so, you can use qsort() to do the
sort, and bsearch() to search through it -- this will be very fast, and
will work with data structures.  If the array is not dynamic, SORTA can be
used to sort it, but it's slower than qsort().  You can still use
bsearch() though.

Another idea would be user indexes...  though, I don't know that I'd go
through all the trouble for a mere 64-150 elements.