[m-users.] uncaught Mercury exception using mdprof_create_feedback

[Paul Bone]

On Wed, Oct 01, 2014 at 09:09:17PM +0200, Matthias Guedemann wrote:
> Dear Paul,
> 
> > The autoparallelisation analysis scans your whole program for places
> > where parallelism could help (it skips parts of the program where it's
> > obvious that it won't help).  It crashes when it looks in
> > euler_81_82_83_problem_set.euler_edge_81/4 which is nondeterminisic.
> > It tries to calculate the cost of every goal including the disjunction
> > as a whole, however it assumes that the disjunction is either a switch
> > or a "semidet disjunction" which is another special case.  This
> > incorrect assumption causes the analysis to crash (because a
> > consistency check fails).  It might be possible to comment out the
> > consistency check, but then I'd be concerned that something else would
> > crash because of the same faulty assumption.  I'll see if I can fix
> > the bug in the next week or so.
> 
> thank you for the explication, I had already supposed this predicate to
> be the culprit. I would be great if the implementation of implicit
> parallelism would support this, too. While it is obvious how to
> parallelize this program manually, I'm very interested in the implicit
> parallelism feature of Mercury, maybe I should have a closer look at
> your thesis ...
> 

The automatic parallelisation analysis tries to be very clever in both it's
search for places to parallelise and how to parallelise them.  However, as
I've worked on it and written my thesis I've found even more situations that
it doesn't work well or at all for; it's one of those situations where you
start to solve the problem but each step you take just helps you realize
that the problem is bigger than you had imagined.

I would say that the thesis is interesting (it kept me busy for 4 years + 2
years for my honours report (part time)).  I've also had independent people
praise the thesis for being well written and straightforward.  (Zoltan and
Peter my supervisors diserve some of this credit too.)  So I do recommend
reading it.

Thanks.


-- 
Paul Bone