[boost] Re: boost::thread construction with unnamed temporary

From: Chris Byrne (chris_at_[hidden])
Date: 2003-10-02 22:55:34

On Thu, 02 Oct 2003 10:01:49 +0200
Daniel Spangenberg <dsp_at_[hidden]> wrote:

> 
> The rule is: "Anything what can be a declaration is a declaration!"
> 
> > Futher more, how can an instantiated variable idx actually be used
> > in a function declaration?
> 
> Function parameter identifier in function declarations never interact
> with other valid names outside
> the scope of the function declaration, because they are local to scope
> of the function prototype. Otherwise you would change the meaning of
> function declarations in situations where you introduce further names.
> 
> Hope that helps,
> 
> Daniel Spangenberg
> 
> 

With a bit more investigation I discovered that:

 boost::thread thrd2(MyThread);
 boost::thread thrd2(MyThread idx);
 boost::thread thrd2(MyThread(idx));

all result in the same definition. So with a bit of a read through Stroustrup I finally came across A.5 which has the very rule you just quoted at me ;-)

"if it could possibly be interpreted as a declaration, it is a declaration"

 char(x); // declaration of a char called x

So that line:
 boost::thread thrd2(MyThread(idx));

pseudo like:
 funcret functypename(parm(parmname));

Which makes sense now, so thanks for that.

The last problem I have however, is should an interface exist where this confusion can easily be made?
Wouldn't it be better to provide a member function to create a thread instead of using the constructor?

- Chris