On Wed, 16 Jan 2013 13:46:03 -0500, David Andrews wrote:
>On Wed, 2013-01-16 at 13:35 -0500, Paul Gilmartin wrote:
>> Consider the Dartmouth/GE time
>> sharing system, which ran on hardware with no memory protection,
>> relying on array bounds checking by the FORTRAN and BASIC processors.
>> Frustrating. To do this they enforced some rules they invented that were
>> never part of the FORTRAN standard.
>
>Namely? (Just curious.)
>
(Suddenly occurred to me that I neglected to answer this.)
ASSIGN 100 TO L
...
GOTO L
Perfectly OK according to the Standard, provided "100" exists as a
label and "L" is an integer variable. But then:
L = 42
... is reported as an incompatible use of L, although allowed by the
Standard. This gives some clue as to what the compiler is doing to
protect the kernel at run time.
-- gil
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
