[Mulgara-general] Performance of disjunction.

[Paul Gearon]

Hi Andrae,

Sorry, I've been busy with SemTech and haven't had a chance to get to
this before now....

On Mon, May 19, 2008 at 3:27 AM, Andrae Muys <andrae at netymon.com> wrote:
> Something that has come up off-list is the performance of disjunction
> in mulgara.
>
> Put bluntly, disjunction is completely unoptimised; and this may be a
> problem for users, but todate we have had very few complaints.
>
> The relevant bug report is http://mulgara.org/trac/ticket/15.  This
> has historically had an implementation focused title, I have updated
> the title to read "Product of Sum form has very poor performance.".
>
> If you are affected by this performance issue, please let us know ---
> and record your interest in the bug-tracker.
>
> To a large extent only squeaky wheels get grease, but this problem
> has been bothering me for sometime, I just haven't been able to
> justify taking the time to fix it.

Actually, I'm regularly telling people to use unions for their graphs,
which turn into disjunctions in the constraint expression.

The rules engine also uses disjunctions (if the inferred data goes
into a separate graph to the original data). To date the rules engine
hasn't had a LOT of use, but it's starting to pick up. Personally,
I've been using it more and more, and Amit tells me that it may be
getting a real workout in the near future.

So yes. I really need this done!

:-)

Paul