[m-rev.] for review: programatically turn profiling on and off

Zoltan Somogyi zs at csse.unimelb.edu.au
Wed Sep 27 12:04:19 AEST 2006


On 26-Sep-2006, Peter Ross <pro at missioncriticalit.com> wrote:
> > >>
> > >>Provided no-one sees problems with this change.  The next plan is
> > >>to add a library module which allows one to turn profiling on and
> > >>off as impure predicates.
> > >>
> > >
> > >I don't think they need to be impure, since they can go inside trace
> > >goals (and therefore have access to the IO state).
> > >
> > >What about adding the new preds to the benchmarking module instead of
> > >adding a new module?
> > >
> > 
> > I don't think that the benchmarking module is the appropriate module
> > to add this to.  I would rather have a new module, e.g. runtime_util,
> > or some thing that included the predicates to turn profiling on or off
> > and the predicate to write out trace counts etc.
> > 
> I will add them in the new module, runtime_util unless there are any
> other objections.

I have an objection to the name; it is far too general. I would prefer
something like "performance_measurement" or an abbreviation.

Zoltan.
--------------------------------------------------------------------------
mercury-reviews mailing list
Post messages to:       mercury-reviews at csse.unimelb.edu.au
Administrative Queries: owner-mercury-reviews at csse.unimelb.edu.au
Subscriptions:          mercury-reviews-request at csse.unimelb.edu.au
--------------------------------------------------------------------------



More information about the reviews mailing list