| Back to Intercal Home Page | Up to Site Map | $Date: 1998/06/06 15:09:59 $ |
While rereading the INTERCAL reference manual, it occurred to me that INTERCAL is a bit dated, since the languages whose features it avoids (Fortran, COBOL, etc.) are pretty much dead. Shouldn't INTERCAL start avoiding features found in some more modern languages, like C++, Java and Perl?
In this vein, let me suggest a direction for future development: overloaded operands. I suggest the following syntax:
<var>/<expr>
<expr1>\<expr2>
In the first form, <var> is a variable (16- or 32-bit, scalar or
array). Subsequently, any time it appears as an operand to $ , ~ , & ,
V , or ? , it will be silently replaced by <expr>. Thus,
DO .7/#&77
DO :8<-.7
will have the same effect as
DO :8<-#&77
In the second form, <expr1> is a 32-bit expression, divided into two
16-bit numbers (high 16 bits and low 16 bite). This defines a range.
Any time a variable whose number is in the range thus defined appears
as an operand, it is replaced with the value of <expr2>.Furthermore, any time the special token ) appears in <expr2>, it is replaced with the number of the variable being affected. Thus,
DO #196608\#99
will affect variables .3 through .10, :3 through :10, ,3 through ,10 ,
and ;3 through ;10 , replacing them with #99 whenever they appear as
an operand to $ , ~ , & , V , or ? .
DO #196608\;)SUB".&)"
DO :1<-.9
has the same effect as
DO :1<-;9SUB".&9"
Since overloading is not indicated by a keyword, there can be
no PLEASE ABSTAIN FROM OVERLOADING directive, but overloading can be
undone by mapping variable numbers back to their original values.There are, of course, various issues that still need to be ironed out, but it seems to me that operand overloading can be a powerful technique, on a par with Perl closures or Java's on-the-fly class definitions.
| Back to INTERCAL Home Page | Up to Site Map | $Date: 1998/06/06 15:09:59 $ |