[Kst] plotting lots of high resolution data
iaindb at netspace.net.au
Sat Aug 21 09:38:33 CEST 2010
I've been playing around with a couple of graphing tools to improve on a
SCADA trending tool I use regularly, but there are some things I haven't
been about to figure out in a couple of hours!
First, how fine can kst handle time values? We plot various types of
data from years of slow temperature values (ie once per second) to
months of fast power / frequency at 100ms or so, to high speed
disturbance data imported from power meters (such as 10 seconds worth of
20ns sampled points). Except for the last example, the rest of the data
is "differentially" recorded not sampled, ie. depending on the rate of
change of the data, the distance between any two data points will be
I've imported an example of about 170,000 samples which look a bit like
this (the data is actually in a binary format but I've converted it to
csv for now):
As a pure x,y scatter plot, kst handles this perfectly! However the x
axis shows labels such as "+[1.0e+05]". When I try and interpret the
data as a time, nothing changes on the x-axis. Probably because it's
down to hundreds of micro-seconds rather than just ms, and I assume this
is too fine for the a kst time?
What format should a ms time be in? I found a post here about ms time:
http://mail.kde.org/pipermail/kst/2008-December/016025.html but I didn't
see any response further than "yes it's possible".
I could get around this and deal with data down to ns or further if
there was a way of putting my own labels on the x-axis. For example, at
"1273262951.5737" I could put a text label of "8/5/2010" or "05:39:11"
depending on the zoom level.
I'm happy to hear your thoughts / suggestions! thanks,
Iain Buchanan <iaindb at netspace dot net dot au>
Historically Tcl has always stored all intermediate results as strings.
(With 8.0 they're rethinking that. Of course, Perl rethought that from
-- Larry Wall in <199710071721.KAA19014 at wall.org>
More information about the Kst