Atomicity of Cells

[Raphael Collet]

David Hopwood wrote:
> 
> Raphael Collet wrote:
> > Dear Christopher,
> >
> >>Exchange is atomic on cells, what about '@' and Cell.access?  I am
> >
> > If C is a cell, then X=@C is the same as X={Access C}, which is the same
> > as {Exchange C X X}.  There is only *one* stateful operation, so it is
> > atomic by definition.
> 
> I don't mean to be pedantic, but the fact that there is only one visible
> operation doesn't necessarily mean that it is implemented in only one
> step. Your argument would also depend on the fact that steps are strictly
> interleaved, which is true of Oz but not some other concurrency models.
> 
> Exchange operations are atomic, but that's a specified property of Exchange
> rather than being true by definition.

You're completely right.  The atomicity of Exchange is guaranteed by the
definition of the language.  This wouldn't be true if Exchange was
defined by

proc {Exchange C Old New}
   Old=@C
   C:=New
end

Cheers,
raph