Re: Compiler and interpreter origins

Dick Weaver wrote:
> There was a language TRAC (Deutsch?) for the PDP-1 where the a program
> could modify itself (I think).


This isn't about TRAC, but (1967 or so) I spent a month trying to
understand the core of the Thor timesharing system for the PDP-1
written in PDP-1 assembler. I understood it all except for one place
where, in the code, in one location either of two data words would be
stored. One word was a "SZA" ("skip on zero accumulator") and the
other was a "SNA" ("skip on non-zero accumulator"). Why would anyone
modify a skip instruction? It was explained to me later that the code
had to be self-modifying. I swore that I would never write
self-modifying code, and I have kept my promise for 37 years. (And I
remember quite clearly who wrote that code, but I bet he doesn't
remember doing it.)


Klaus Wirth used to go off on self-modifying code; he
considered it a terrible evil.


john slimick
slimick@pitt.edu
[I missed the reference to Trac, a language in which I've probably
written more code than anyone. It's a strictly interpretive macro
expansion language somewhat similar to Unix m4, so just about anything
you write is by some standard self-modifying code. The code I wrote
didn't do much self-modifying beyond putting commas into arguments to
recursive routines to implement variable length argument lists and
poor man's structures. -John]

Return to the comp.compilers page.
Search the comp.compilers archives again.