Re: Operationalize orthogonality
From: Tony D <tonyisyourpal_at_netscape.net>
Date: 1 Jun 2006 14:37:49 -0700
Message-ID: <1149197869.175138.27780_at_i39g2000cwa.googlegroups.com>
Date: 1 Jun 2006 14:37:49 -0700
Message-ID: <1149197869.175138.27780_at_i39g2000cwa.googlegroups.com>
My favourite "unfortunate series of unforeseen interactions" is still the old FORTRAN DO-loop screw up that managed to lose a Viking Venus probe, apparently.
Consider the meaning of the FORTRAN sequence
DO 20 I = 1, 100
* stuff here
20 CONTINUE
Now consider the meaning of the following sequence
DO 20 I = 1. 100
* stuff here
20 CONTINUE
Now consider the following "features" of FORTRAN.
- White space is not significant
- There are no reserved words
- Variables are automatically declared for you on first use if you didn't bother explicitly declaring them
- An automatically declared variable beginning with letters other than I,J,K,L,M or N is assumed to be of type REAL
- A labelled CONTINUE statement that isn't addressed by a DO-loop is treated as a no-op
Now consider the proximity of , and . on the keyboard.
Whoopsie.
Gene Wirchenko wrote: [ with another excellent reason why PL/I should fall off the face of the earth ] Received on Thu Jun 01 2006 - 23:37:49 CEST