[Date Prev][Date Next][Thread Prev][Thread Next][Author Index][Date Index][Thread Index]

Re: No more patches



> On Fri, Oct 30, 1998 at 04:47:26PM -0500, Mark-Jason Dominus wrote:
> > So I'm going to hold off on making any more changes until one of the
> > following occurs:
> > 
> > 1. Andrew gets the CVS working, or
> > 2. A committment is made to include some or all of my patches and a new
> >    version is released that includes them, or
> > 3. All my patches are officially rejected.
> 
> I've included both of your bugfixes and your massive patch (with some
> rewriting, notably the atrocious get_links_from() replaced by a new
> get_links_to() function which I'm going to give you the enjoyment of
> integrating with your copying code *grin*) 

Fine, that is a better way to do it, and I'm glad you got rid of
get_links_from, which was disgusting.

But I think my original point about the misdesign stands: There is no
good way to find out what cells are linked from a certain cell.  Your
suggestion was to enumate all the dimensions and then search in every
possible dimension for links.  That is not as bad as what I was doing,
but it is still bad.  I imagine that if ZigZag is going to be useful,
the typical space will have at least a few hundred dimensions.  It is
silly to have to resort to brute-force search for something so
fundamental. 

I like Bek's idea of a three-column table much better.  If searching
the table is efficient, then all the typical queries that we need to
make will also be efficient.  Methods for making fast queries on this
table are already well-known.