Re: [boost] Re: boost.threads

From: Greg Colvin (gcolvin_at_[hidden])
Date: 2001-08-14 12:06:21

From: "John Max Skaller" <skaller_at_[hidden]>
> Greg Colvin wrote:
> 
> > My thinking is just that since a compiler can easily take
> > advantage of a fast local stack without banning shared stack
> > data, just as it does for registers, there is no compelling
> > need to inconvenience programmers.
> 
> Lets kill this issue: it died immediately Beman said he shared
> stack data. The only way a bad on sharing stack data would 
> be viable is if hardly anyone ever did it.  Excuse my ignorance.

For me the issue has changed.  It is a given in C and C++ that you
can take the address of an auto variable.  Any language feature
that can be used will be used, and abused, so we can't just forbid
this in multi-threaded code.  The question now is, given the kinds
of hardware, OS, and compiler support that exist:  What is the
performance impact of forbidding shared stack data?

I conjecture that is needn't be enough to worry about.