Stu Cram wrote:

> Just a few questions about some variable types in FB5 as I'm not sure how accurate the HELP is about them.
> 
> 1.  	I used DIM AS INTEGER and the program worked as expected.
> 	I also used DIM as INT and it still worked.
> 
> 	However, INTEGER is not listed in the HELP file as a variable type but INT is.
> 	INT is also a function to round a real to a LONGINT value.
> 	
> 	Can you see some possible confusion here? 
> 	Are INT and INTEGER synonyms wrt to DIM AS?

These are equivalent ways of declaring a 16-bit signed integer variable:
dim as SInt16  x
dim as short   x
dim as int     x
dim as integer x


> 2.	REAL cannot be used as a variable type apparently; I thought it might be a synonym for SHORT.
> 	Wish it were so. Or perhaps a synonym for DOUBLE, that would be better.
> 	Yes I know I could add #DEFINE REAL AS DOUBLE, but I would rather have that built-in for everyday simplicity if possible.

Floating point types are single and double. 'real' has no meaning to FB unless you #define it.

> 3.	DIM AS FIXED is another valid real type variable (with a limited values range).
> 	However, the keyword FIXED does not appear in bold in a DIM AS statement like other built-in variable types.
> 	Is that an oversight in the editor? Maybe FIXED is missing from a keyword list? Just a thought.

The file "Key Words" in the Headers folder specifies what the editor should display in keyword style, and "Fixed" is missing from that file. Added for the next release.


> 4.	Way back in FB2, I think there was an option to set the precision for fixed variables so that very long numbers could be used. Has there been thought to bring that back. It seemed very natural to me at the time to specify 50 or even 200 decimal places for example to work in my math classes when showing repeating decimals, value of π, etc.  REAL and DOUBLE values just don't show enough decimal places.  Sigh, where are the 'good old days'?

The (software-based) floating point arithmetic in FBII was idiosyncratic and extremely slow.

Robert P.