Re: [Gpsbabel-misc] position filter bug in gpsbabel 1.3.3?

[Vidar G. <vid...@37...>]

===== Original message from Olaf Klein | 27 Feb 2007:
> ".... A point is removed if it is within the specified
> distance of a point that has come before"
>
> Your trip comes back on the same way to the start
> point. So a lot of these points are dropped because they
> are in the near of points from the first half.

exactly, and my point is that
this is not what i expect the filter to do.

i expect the position filter to filter multiple points in
the same position, but not when this position are returned
to at a later point. so i disagree with the current
solution, and think it produce erroneous output.

let me explain by example:
if i merge a lot of tracks recorded from trips in the same
area over a period of time, e.g. mountain bike trips last
summer or all cross country skiing trips this
summer. obviously, the same tracks must have been followed
several times, but this is not redundant samples, as the
1.3.3 position filter regards it.


an example of this is this Google Earth file,

 http://37mm.no/ge/kjekstadmarka.kmz

which i could want produce using a python script

 e.g. ./gplQuick.py kjekstadmarka*.gpl

which merges all track logs, and splits again at 15min gaps.
this would work fine* with GPSBabel 1.3.2, but not with the
changes made to the position filter in GPSBabel 1.3.3.

*although i agree that the position filter in 1.3.2 had it's
flaws: it failed to filter all samples near one position if
the GPS were basically in one spot for a long time.


i realize that both uses of the position filter could be
relevant, so i'm merely pointing at the trouble with it.
one solution could be to offer an extra option to set a
time/distance horizon, e.g. position,distance=20m,within=10min


Vidar___