[geos-devel] RE: LineSegment copy removal, ABI protection 2

[Martin Davis]

> Martin, are you experiencing this much improvements ?

No.  But I don't think my recent changes would be executed during buffer
anyway.  And I certainly wouldn't expect them to provide that much
improvement.

Let me get this straight - the changes that you made were just to change
LineSegment to use references rather than copies?  (This change is
irrelevant in the Java world, since the usual pattern is to use
copy-by-reference.  So JTS effectively already has this improvement, I
think.)

Martin Davis, Senior Technical Architect
Vivid Solutions Inc.      www.vividsolutions.com
Suite #1A-2328 Government Street Victoria, B.C. V8T 5G5
Phone: (250) 385 6040 - Local 308 Fax: (250) 385 6046


> -----Original Message-----
> From: strk at refractions.net [mailto:strk at refractions.net] 
> Sent: June 27, 2005 6:43 PM
> To: GEOS Development List
> Cc: Martin Davis
> Subject: Re: [geos-devel] RE: LineSegment copy removal, ABI 
> protection 2
> 
> 
> On Mon, Jun 27, 2005 at 11:20:04PM +0200, 
> strk at refractions.net wrote: ...
> > Ok, first of all I'm fried, sorry. My timings were distorder by the 
> > profiler. W/out specific profiling calls comparison
> > is:
> > 
> > With LineSegment::reverse calls
> >         real    0m35.149s
> >         user    0m30.244s
> >         sys     0m1.231s
> > 
> > Without LineSegment::reverse calls
> >         real    0m29.858s
> >         user    0m29.043s
> >         sys     0m0.121s
> 
> I might be fried again, but after porting latest Martin's 
> patches time sems to be reduced to:
> 
> 	real    0m8.847s
> 	user    0m8.542s
> 	sys     0m0.093s
> 
> Martin, are you experiencing this much improvements ?
> 
> --strk;
> 
> > XML test is available from: 
> > http://foo.keybit.net/~strk/tmp/test.xml.bz2
>