FB - DIM REGISTER vs VAR

Mickster

Which do I choose for local variables and why?

Craig

ersmith

DIM REGISTER declares a global variable that lives in a register. This is generally a niche thing to do. The normal way to declare local variables is with just plain DIM or with VAR. The only difference is that VAR requires you to initialize the variable, and infers the type of the variable from the type of the expression. That is inside a function

VAR x = "hello"

is equivalent to

DIM x AS STRING
x = "hello"

Outside of any functions they are slightly different, with VAR declaring a variable that's not accessible to subroutines and functions (i.e. that's "local" to the PROGRAM statements).

Mickster

Thanks @ersmith Eric.

I switched from a bunch of globals to VARs and the speedup was shocking

What had me confused was that I got the impression that; whatever the VAR is initialized with, establishes its type.
For example:

Var a = 1

A ubyte would suffice but I actually need a signed 32bit integer. Well, I actually do end-up with what I need so I guess that what the text really means is that:

Var a = "1"

Establishes the variable as a string

Craig

ersmith

@Mickster said:
Thanks @ersmith Eric.

I switched from a bunch of globals to VARs and the speedup was shocking

What had me confused was that I got the impression that; whatever the VAR is initialized with, establishes its type.
For example:

Var a = 1

A ubyte would suffice but I actually need a signed 32bit integer. Well, I actually do end-up with what I need so I guess that what the text really means is that:

Var a = "1"

Establishes the variable as a string

Byte and word operations are slower and take up the same number of registers as long, so all integer expressions are treated as 32 bits for purposes of VAR. But there are more types than just string and integer, e.g.:

VAR a = 1  ' a is an integer
VAR a = 1.0 ' a is a single
VAR a = [x : => x+1]  '  a is a function
VAR a = b  ' makes a have the same type as b, *whatever* that is

Mickster

@ersmith said:

Byte and word operations are slower

Ah, this I already expected to be the case but rather than appear lazy, was going to test.

Thanks again.

Craig