Re: No exceptions?

From: Marshall <>
Date: 30 Jun 2006 08:01:45 -0700
Message-ID: <>

paul c wrote:


> However, the interpretation of a language could be defined such that the
> mention of "x" without any collateral description stands for a relation
> with no attributes.

One could certainly do this, but I believe it would turn out to be a bad idea in practice. Every typo of an identifier would still make a valid program, just not the one you thought it was.

You ask about exceptions, which are runtime constructs, but you example here is about compile time constructs. Syntax errors aren't exceptions in the usual sense; they indicate that a stream of characters is not a program.

Consider that at the endpoint, it would be possible to design a language such that every stream of characters was a valid program; this would *not* be a desirable property. (But I do agree that reducing exceptions is a desirable goal.)

Marshall Received on Fri Jun 30 2006 - 17:01:45 CEST

Original text of this message