Correction to the DIM statement for BASIC 4r32.
HIBASIC 4.32 added the following check in the Multiply Upper Dimension by Lower Dimension routine:
[&95E4] BCS 11 --> &95F1 Bad Dim error
this occurs after we have multiplied &2A-&2B by 2, and avoids memory corruption if the number of elements to dimension (&2A-&2B) is too large. This check now correctly disallows arrays to be DIMmensioned, where the number of elements would exceed 65535 (&FFFF). In BASIC IV, DIM K(1582,127) was allowed (total of 200914 elements!) which caused the following weird bug:
>L. 10 DIM K(1582,127) 20 FOR I% =1 TO 1582 30 FOR J%=1 TO 127 40 K(I%,J%)=56.77777 50 NEXT J% 60 NEXT I% >RUN Array at line 40 >L. 0 DIM K(1582,127) 0 FOR I% =1 TO 1582 0 FOR J%=1 TO 127 0 K(I%,J%)=56.77777 0 NEXT J% 0 NEXT I% >
Memory was corrupted, causing BASIC to become confused. In BASIC4 r32, DIM K(1582,127) now produces 'Bad DIM' error.