[OSM-dev] Osmosis Plans
jburgess777 at googlemail.com
Tue Sep 25 19:30:46 BST 2007
On Tue, 2007-09-25 at 10:45 +0200, Martijn van Oosterhout wrote:
> On 9/25/07, Brett Henderson <brett at bretth.com> wrote:
> > 2. Again, Jon Burgess has written an excellent C implementation of a
> > planet difference generator that is somewhat faster than osmosis (mainly
> > due to osmosis parsing data into full objects incurring a large date
> > parsing overhead and writing data back into string format). Again,
> > osmosis is not necessary here.
> To be honest, the fact that you do normal parsing is *good*. I think
> the time I wasted due to regex parsing producing the wrong answers is
> probably more than the time saved by such an implementation. As the
> saying goes: I can give the wrong answer in no time at all.
To clear this up. Planetdiff certainly does not perform regex based
parsing. It uses a proper libxml2 based parser. The key difference is
that the planet diff too treats the contents of the timestamp field as
an opaque string. It makes no attempt to parse this into a C data type.
It determines whether the timestamp has changed by using a plain string
Osmosis on the other hand treats it as a real data field so if it
changed format but the 'value' stayed the same then it'd ignore the
differences (e.g. "2007-09-25 12:00:00" vs "2007/09/25 12:00:00").
More information about the dev