Determining Variable Type

[MattC]

Hi.

Is there a way to determine (at run time) the type of variable that is being used, i.e. whether the variable, array, etc. is an integer, floating, string, etc.?

Matt

[hellomike]

Hi,

Maybe with a combination using the ^ operator and the "Variable storage in memory" help section....

Did you try/test that?

Mike

[MattC]

Thanks.

I've tried a variety of ways, but all of them need to know the variable before its type is determined.

The reason I wanted to know was in order to create a subroutine that dimensioned an array from DATA elements, but that the routine determined the type of array that needed dimensioning. That is, you could pass any type of array to the single routine (directly or indirectly) and the routine would do the rest. I've got a basic one working, but it has four DEF headers, each knowing ahead of time which type the array is. (Not sure how well I explained that.)

Code: Select all

      REM Returns a dimensioned array and filled with the data listed DATA.
      REM *** NOTE *** The original array passed by reference must be a GLOBAL array
      REM              and not one assigned by LOCAL.
      REM              It is important to use the correct PROC's name.
      REM *** NOTE *** Requires the last DATA element to be "END_DATA_SET".
      REM              This element will NOT be saved to the array.
      DEF PROCDataFillBytArray(RETURN array&(), datapoint%) : LOCAL T% : T% = 0
      DEF PROCDataFillIntArray(RETURN array%(), datapoint%) : LOCAL T% : T% = 1
      DEF PROCDataFillNumArray(RETURN array(),  datapoint%) : LOCAL T% : T% = 2
      DEF PROCDataFillStrArray(RETURN array$(), datapoint%) : LOCAL T% : T% = 3
      LOCAL I%, C%, W&, W%, W, W$
      RESTORE datapoint%
      REPEAT
        READ W$
        C% += 1
      UNTIL W$ = "END_DATA_SET"
      C% -= 2
      IF C% = -1 THEN ENDPROC
      CASE T% OF
        WHEN 0: DIM array&(C%)
        WHEN 1: DIM array%(C%)
        WHEN 2: DIM array(C%)
        WHEN 3: DIM array$(C%)
      ENDCASE
      RESTORE datapoint%
      FOR I% = 0 TO C%
        READ W$
        CASE T% OF
          WHEN 0: array&(I%) = VAL(W$)
          WHEN 1: array%(I%) = VAL(W$)
          WHEN 2: array(I%)  = VAL(W$)
          WHEN 3: array$(I%) = W$
        ENDCASE
      NEXT
      ENDPROC

The code requires data to be placed after the appropriate label such as (data_list) and finishes with the final element of "END_DATA_SET".

The array is only one dimensional, but it still works. I'll expand it further if I need it. However, I don't like the four DEFs, and there are so many CASEs that it looks like a lost luggage department.

Matt

[hellomike]

Yes, I see. Looks tricky.
Seems you can get rid of all local variables starting with W, except W$, not?

Anyhow, can't think of a more straightforward way of achieving what you need.

Would be nice if Richard would still be on-board to give advise.

Mike

[MattC]

I've been using the Cross Reference utility and that obviously has a way of determining what type of variable is being used. So I guess there must be a way - unless it determines it directly from the listing.

I agree with you final comment about Richard.

Matt

[hellomike]

Cross Reference only looks at the tokenized source. Program isn't running at that moment.

Pretty sure it interprets the source about the same as interpreter does during run time. So it simply uses the actual last character of the variable name. %, $, &, <none>. Their purpose is not just readability I'm sure.

Have you checked the help for the CALL statement. That does a lot with parameter types!

Also related is:

Calling internal routines from assembler code

Hope this helps.

[hellomike]

this wiki page states:


CALL @fn%(4) ; Look up variable (esi points to name, returns type=al, varptr=ebp)


........

[MattC]

Thanks. I'll have to look into the CALL statement.

I don't have v6 so I can't use the @fn%() function.

Thanks for looking.

Matt