[m-dev.] Thinking about mode analysis and uniqueness

[Fergus Henderson]

On 03-Jan-2003, Ralph Becket <rafe at cs.mu.OZ.AU> wrote:
> Fergus Henderson, Friday,  3 January 2003:
> > I presume these are all definite properties (i.e. X ~ Y means that X
> > definitely aliases Y, and X < Y means that Y definitely contains X)?
> 
> They may be conservative approximations.

Sure, but the question is what do you mean by "conservative"?

A "conservative" approximation for sharing is not necessarily
the same as a "conservative" approximation for aliasing --
in the former case, you should over-approximate (i.e. analyze
*possible* sharing), since the optimization which you're trying
to apply depends on *not* sharing,
but in the latter case you should under-approximate (i.e. analyze
*definite* aliasing), since the optimization that will be applied --
assuming that the inst of one variable is the same as the inst of an
aliased variable -- requires the aliasing property to hold, rather
than requiring it to not hold.

So for each run-time property which you want to "conservatively"
approximate, you should specify whether you want to analyze whether the
run-time property definitely holds, or just whether it possibly holds.

-- 
Fergus Henderson <fjh at cs.mu.oz.au>  |  "I have always known that the pursuit
The University of Melbourne         |  of excellence is a lethal habit"
WWW: <http://www.cs.mu.oz.au/~fjh>  |     -- the last words of T. S. Garp.
--------------------------------------------------------------------------
mercury-developers mailing list
Post messages to:       mercury-developers at cs.mu.oz.au
Administrative Queries: owner-mercury-developers at cs.mu.oz.au
Subscriptions:          mercury-developers-request at cs.mu.oz.au
--------------------------------------------------------------------------