[Mulgara-general] Re: [Mulgara-dev] Removing exclude.

[Andrew Newman]

On 4/14/07, David Wood <dwood at softwarememetics.com> wrote:
> On 12 Apr2007, at 23:58, Andrae Muys wrote:
> > Do we agree that exclude should be removed?
>
> I have no objection.
>
> Andrew's point about SPARQL's NAF should be addressed prior to a
> SPARQL implementation, but I do not object to a removal of exclude,
> clean up and then reimplementation as required.
>

I can understand getting rid of something that's actively confusing to
people to use.  So getting rid of the grammar and tuples operation is
one thing - no big deal as minus covers that with a better syntax and
a better implementation (exclude does a scan of a much larger part of
the store compared to minus).

So there's SPARQL and as I've expressed before, NAF, especially in
custom rules (non-OWL) is useful.  Loosing this would be bad for this
reason too.

Maybe it people just want to reimplement it a different way?

> > If it should be removed, when should this occur?
>
> If we cannot find a customer using it, I support removal at the next
> release.  Our user base is arguably small enough for that.  The
> presence of minus and suggestions for migration of existing code MUST
> accompany the first release without it.
>

BTW, I agree that if no one is using it and it's causing grief for
maintenance that it should be removed.  I would've thought it'd be
easier just to document minus and explain that exclude should hardly
ever be used because it doesn't do what most people think it does.