[geos-devel] GEOS<3>

[strk at refractions.net]

On Wed, Apr 20, 2005 at 09:30:27AM +0100, Artem Pavlenko wrote:
> strk at refractions.net wrote:
> 
> >On Tue, Apr 19, 2005 at 04:52:12PM -0400, Mark Coletti wrote:
> > 
> >
> >>On 4/19/05, Artem Pavlenko <artem at pavlenko.uklinux.net> wrote:
> >>   
> >>
> >...
> > 
> >
> >>>Meanwhile , could anyone explain to me why do we need
> >>>vector<Coordinate>*  and not just vector<Coordinate>
> >>>in DefaultCoordinateSequence? 
> >>>     
> >>>
> >>Oh, my.  I hope it's not for some perverse form of vector re-sizing.
> >>   
> >>
> >
> >I found it like that and failed to clean it up as most of the current
> >code. Changing - again - will break API.
> >Actually I think we can really start a GEOS3 branch and do a big
> >lifting on it.
> > 
> >
> I'm up for it:-)
> 
> Have you heard anything from Dale Lutz?

He announced is going to send me the patches, some of which
are controversial (I think the CoordinateSequence ones, non-allowing
higher-dimensioned Coordinates) but compile-time definable.

For the GEOS<3> branch I'd first make a plan, as I belive the hard
thing with GEOS is keeping a stable ABI. The first change would be
some kind of ABI wrapper or similar thing ... but I don't know yet
how to do it. The only thing I imagined was a flat-namespaced C wrapper
like the one produced for JTS and the one implemented inside postgis.

I'm currently working on porting the WKB parser for specifically this
purpose.

> >>And again, why  not write:
> >>   
> >>
> >>>  struct A
> >>> {
> >>>     A(B const& b)
> >>>       : b_(b) {}
> >>>      B b_;
> >>> };
> >>>     
> >>>
> >
> >We can do this and I actually started doing it somewhere, but as far
> >as args are pointers this doesn't really speed things up...
> > 
> >
> True, but it is confusing. Lets make GEOS better platform!

Patches in this direction are accepted.

--strk;

> 
> Cheers
> Artem
> _______________________________________________
> geos-devel mailing list
> geos-devel at geos.refractions.net
> http://geos.refractions.net/mailman/listinfo/geos-devel