[snipped long claim against Blitz's apparent limitation to 21 Million as
 maximum value storable in a variable]

What about starting a REAL programming-work by changing your number
representation ? You should calm down and start understanding computers
arithmetics before giving so strong judgements !
First of all Blitz has no maximum limits on number's size, but the
supported standard types have, so, you are using a long type variable
you'll have a limit, only because
your number will be stored with 4 bytes and considered to be 
a signed integer. With such a number representation EVERY computer
will have the 21 Mio limit !!
But NOONE has ever sayd "in Blitz 
the maximum number size can be achieved by a variable of the
long type", just pass to the float type and this limit
will increase a lot. Anyway this is not the perfect solution,
because of the loss of accuracy, but this will not mean that
Blitz itself cannot store numbers of any NEW type that
you can define putting into it very BIG sizes and accuracy.
Now, try to get a wider view and create a representation for your numbers
wich stores (for example) the values in 8 bytes considering
them as unsigned integers : you'll get a maximum value of 2^64-1, wich
should be enough (if not then let pass to 16 bytes and you'll get a
max size large enough to store the number of 50th of second passed 
since the start of the world :).
Naturally, to manage such a new representation is harder than simply
using the pre-defined long type with the pre-defined (automatically
managed) long-type arithmetic (this will mean to define procedure
wich will do the addings, muls, subs, and so on applyed to 8-16 bytes
instead of only 4) ... but it's more than possible ! And please
don't say me that in other languages this "bignumbers" arithmetic 
is already defined, and because of this blitz is s*it ! In that "other
languages" other important things (like automatic string handling,
I mean that you can add with the operator "+" strings
as they were numbers, wich
is implemented in Blitz but not in C ) are supported and the programmer
has to find a representation for their new datatypes and develop some
procedures to work with them, exactly what you should do for your
finacial program.