[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Why would anyone want opacity?

From: "Guillermo J. Rozas" <gjr@martigny.ai.mit.edu>
Subject: Re: Why would anyone want opacity?
Date: Tue, 30 Apr 96 19:07:28 -0400

> So you don't believe in debugging?  The point that I believe you've
> missed is that the _same_ information that is left behind for
> debugging can be used by _other_ meta-programming tools.  In fact,
> once you make that observation, you realize that the debugger is
> _nothing special_.  Users can write their own if they so wish.

I did not miss this point, and, indeed, I do not believe (very much)
in debugging (at least not the kind that is usually referred to as

> Thus saying that the extra work is not worth it is essentially saying
> that debugging is not worth it, and I doubt you believe that.

I hardly ever use a debugger myself, whether I program in C, Scheme,
or SML.  I would never compromise agressive optimization for the
benefit of a debugger, nor would I jump through infinite hoops to make
a debugger `work' despite the efforts of an optimizer.  For me, this
is simply not worth the trouble.

I have no problem with the idea of separate debugging environments,
where the compiler either spends less effort on obfuscating the
original code, or where debugging works by source-code instrumentation.
Unfortunately, both techniques do not work well with less than
completely defined semantics, because different optimization levels or
program instrumentation can (and most often do) change the behavior.

In any case, I do not want to make the debugging environment the
default for a language's implementation.