[m-dev.] Trailing

[Fergus Henderson]

Bart Demoen wrote:
> 
> another point: if functions and "regular" trailings are pushed on the
> same stack, and executed/undone in reverse order of pushing, there is
> possibly a different behaviour from your schema or the 2 stacks schema
> in which this order is not respected; is that acceptable ?

Well, it might be.  But it would be nicer if we could avoid that.

> > So the cost tradeoff is that trailing a value becomes more expensive (a
> > conditional branch which will probably usually be taken
> 
> you have shifted some of the overhead of having tagged entries on one
> trail stack from untrailing to trailing - is it really better ?

I agree with Bart's comments here.  Given also the disadvantage of this
scheme with respect to untrail ordering, I'd be inclined to prefer a
tagged trail stack.

-- 
Fergus Henderson <fjh at cs.mu.oz.au>   |  "I have always known that the pursuit
WWW: <http://www.cs.mu.oz.au/~fjh>   |  of excellence is a lethal habit"
PGP: finger fjh at 128.250.37.3         |     -- the last words of T. S. Garp.