Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!samsung!usc!apple!uokmax!occrsh!fang!att!westmark!mole-end!mat
From: mat@mole-end.UUCP (Mark A Terribile)
Newsgroups: comp.lang.c++
Subject: Re: conditional scope and destructors results
Summary: Yes, but how good is good?  How high is up?
Message-ID: <449@mole-end.UUCP>
Date: 1 Dec 90 08:36:11 GMT
References: <1990Nov13.114958.20853@hemel.bull.co.uk> <329@tin.crc.ac.uk> <1990Nov28.191327.8778@Neon.Stanford.EDU>
Organization: mole-end--private system.  admin: mole-end!newtnews
Lines: 42

> > [Ellis and Stroustrup] say it is illegal to jump forward past an 
> > initializer -- not that it is illegal to write code that might permit
> > a jump past an initializer.
	. . .
> > I would suspect that smart compilers would cause a compiler time error
> > if it can be assuredly determined that a constructor is being by-passed ...

> Maybe this is impossible to do in general, but it should be possible to
> use a conservative strategy. For example, cfront 2.0 throws out the following

> int i=1;
> if (i!=1)  goto L123;
> float x=0.0;
> L123:

> even though the goto will never be executed. There's a case for tightening
> up the wording, but I believe the responsibility for detecting potential
> errors of this kind should rest with the compiler, even if it occasionally
> results in minor inconvenience (through the adoption of a conservative
> strategy).

There's a real problem with this real nice idea, to wit, what defines
the algorithm that determines whether or not the compiler can `assuredly
determine' that the declaration will never be elaborated?

What strict, unambiguous, standard-suitable definition is to be found for
the criteria against which a piece of code is to be tested?

Will such a definition be USEFUL to the ordinary Sue, Ann, and Cheryl
who are programming in C++?

Would it be useful to the maintenance programmer who inadvertantly broke
a flow-control assumption on which depended the correctness of some remote
piece of code?

Some of the answers might be acceptible if there were a well enough developed
body of computing theory.  In fact, little or none of our present computing
theories can give us a good language with which to talk about this problem.
-- 

 (This man's opinions are his own.)
 From mole-end				Mark Terribile


Brought to you by Super Global Mega Corp .com