Path: utzoo!mnetor!uunet!seismo!sundc!pitstop!sun!decwrl!hplabs!otter!kers
From: kers@otter.hple.hp.com (Christopher Dollin)
Newsgroups: comp.arch
Subject: Re: self-modifying code
Message-ID: <780003@otter.hple.hp.com>
Date: 18 Feb 88 15:55:29 GMT
References: <486@taux01.UUCP>
Organization: Hewlett-Packard Laboratories, Bristol, UK.
Lines: 32

"yuval@taux01.UUCP (Gideon Yuval)" says:

> Is good support for self-modifying code a real  issue?  all  CPUs  support  a
> "modify  code  -- invalidate all caches -- execute" cycle, which is enough to
> run  (say)  loaders &  debuggers;  i  THINK  this  is  enough  for  all  real
> applications, but want to be sure.

That depends. In the Poplog environment, for example, user code (Pop11, Prolog,
ML, Common Lisp) is compiled to native code by an incremental compiler. You
could call this "loading", but I suspect that many people will think of a
loader as being something that loads rather larger peices of code than
individual procedures.

What's more, the Pop11 "partial application" feature, whereby a procedure is
constructed from a base procedure by supplying some of its arguments, also
involves the creation of new procedure records in the store. While the compiler
is not often invoked repeatedly in user code, partial application (and to a 
lesser extend procedure composition) is. This means that code cache flushing 
might happen rather often.

Strictly, none of this is "self-modifying" code, since the procedures are
constructed as data-objects before being used. But the data cache has to be
flushed first. 

Of course, some peiple might say that a mixed-language incremental development
system with integrated editor wasn't a "real" application.....................




Regards,
Kers                                    | "Why Lisp if you can talk Poperly?"