Hi Ben,

while it doesn't make too much sense to respond to your questions that
late, here are some thoughts about the two points of your posting:

On 11.12.06, ben lasscock wrote:
> Hi pyx,
> I have made two changes to my copy of pyx.
> (1) I've made the number of significant figures on the tick labels the same
> by padding with zeros.
> So on my figure the tick labels on the vertical axis were:
> 0.8 1.2 1.6 2  2.4 2.8 3.2
> With the change the 2 now reads 2.0. Similarly on the horizontal axis.

Fine. Note that the decimal texter coming with PyX do support this
feature from the very beginning. Take a look at the equalprecision
flag. I know it's a bit hard to spot at the beginning, but you can do
that with PyX without hassards and as it's intended by its creators.

> (2) With a log plot, if the size of the error bar is greater than the size
> of the
> data point then the function  axis.convert returns nan's which go into the
> array sharedata.vrange. This causes the error bar not to be displayed,
> which looks a bit funny. So I've changed it in this case to display the
> error bar to the bottom of the vertical axis. This produces figures similar
> to gnuplot. I find it easier on the eye to have the  rubbish data with huge
> error bars.

It's fully intended that PyX does not show the errorbars for such a
case. Really, it is a feature! To my understanding it's the only right
thing to do here. If you display an error bar on a log scale even with
the errorbar cap outside of the graph range, you still claim the error
bar cap has a positive value, since that is all a logarithmic axis can
express. And that's simply not true. It's wrong, wrong, wrong. You
can't show an errorbor for such an case ... no way. Never.

Being a physicist I would think again: If your problem coordinate is
always positive by definition (say it's a temperature) and you
calculate errorbars by statistical means, you may want to taking the
log of the problem coordinate and do statistics on that. Something
like that might be the only right thing to do anyway. But in case you
can have negative values in your problem coordinate (say for some rare
events), it's only fair to show errorbars on a scale, where negative
values can be expressed. (You should even do so, when statistics
returns positive values only.) Really.


André

-- 
by  _ _      _    Dr. André Wobst
   / \ \    / )   [EMAIL PROTECTED], http://www.wobsta.de/
  / _ \ \/\/ /    PyX - High quality PostScript and PDF figures
 (_/ \_)_/\_/     with Python & TeX: visit http://pyx.sourceforge.net/

-------------------------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems?  Stop.
Now Search log events and configuration files using AJAX and a browser.
Download your FREE copy of Splunk now >>  http://get.splunk.com/
_______________________________________________
PyX-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/pyx-devel

Reply via email to