Bug 272580 - Wrong point in PSDs with log scale
Summary: Wrong point in PSDs with log scale
Status: RESOLVED FIXED
Alias: None
Product: kst
Classification: Applications
Component: general (show other bugs)
Version: 2.0.3
Platform: Compiled Sources Linux
: NOR normal
Target Milestone: 2.0.4
Assignee: kst
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2011-05-06 00:03 UTC by Nicolas Brisset
Modified: 2011-06-30 02:53 UTC (History)
2 users (show)

See Also:
Latest Commit:
Version Fixed In:
Sentry Crash Report:


Attachments
Test case (for PSD options) (11.46 KB, application/octet-stream)
2011-05-06 00:03 UTC, Nicolas Brisset
Details
PSD options I use (58.13 KB, image/png)
2011-06-15 20:53 UTC, Nicolas Brisset
Details
Here is the result I get just after setting the X axis to log and hitting Apply (88.53 KB, image/png)
2011-06-15 20:56 UTC, Nicolas Brisset
Details

Note You need to log in before you can comment on or make changes to this bug.
Description Nicolas Brisset 2011-05-06 00:03:57 UTC
Created attachment 59669 [details]
Test case (for PSD options)

Version:           2.0.3
OS:                Linux

When setting the X axis to logarithmic, there is a point around 1e-281 that looks very strange.

Reproducible: Always

Steps to Reproduce:
- Load CStk from TWIN1xxx, creating XY and PSD in different tabs (PSD options: see attached test case)
- In the second tab, with the PSD, set the scale to log

Actual Results:  
You get a point below 1e-281 which distorts the plot completely

Expected Results:  
Changing the X axis scale from log to linear should not have such a drastic effect!

OS: all
Interestingly, loading the test case does not produce the problem!!! You have to go through the steps above to evidence it.
Comment 1 Netterfield 2011-05-26 03:50:22 UTC
I can't reproduce this.  Can you check again?
Comment 2 Nicolas Brisset 2011-06-13 21:29:01 UTC
Yes, I can confirm the issue is still here. I'm on OpenSuse 11.2 32 bits. Are you on 64 bits?
Comment 3 Netterfield 2011-06-13 21:44:01 UTC
yes.  64 bit.

Can you reliably make this happen?  If so, how>

On Mon, Jun 13, 2011 at 3:29 PM, Nicolas Brisset
<nicolas.brisset@eurocopter.com> wrote:
> https://bugs.kde.org/show_bug.cgi?id=272580
>
>
>
>
>
> --- Comment #2 from Nicolas Brisset <nicolas brisset eurocopter com>  2011-06-13 21:29:01 ---
> Yes, I can confirm the issue is still here. I'm on OpenSuse 11.2 32 bits. Are
> you on 64 bits?
>
> --
> Configure bugmail: https://bugs.kde.org/userprefs.cgi?tab=email
> ------- You are receiving this mail because: -------
> You are the assignee for the bug.
> _______________________________________________
> Kst mailing list
> Kst@kde.org
> https://mail.kde.org/mailman/listinfo/kst
>
Comment 4 Nicolas Brisset 2011-06-15 20:53:56 UTC
Created attachment 61032 [details]
PSD options I use
Comment 5 Nicolas Brisset 2011-06-15 20:54:16 UTC
Now, that's strange! I have reproduced it on a 64 bits OpenSuse Linux machine, so it seems it's not related to that. And it also happens on Widows Xp 32 bits. 
What's more, when I look at the values (View->Vectors) the first frequency is shown as 0.
I am surprised you can't reproduce the problem, as the steps described above lead to it each time. Here are the PSD options again as screenshot, but you can also use kst from an account where it has never been used, as the default values also lead to the issue.

If you really can't reproduce the problem, maybe you can give me hints at where I should look, I may debug it myself. I have actually tried a bit, but I'm not very efficient.
Comment 6 Nicolas Brisset 2011-06-15 20:56:42 UTC
Created attachment 61033 [details]
Here is the result I get just after setting the X axis to log and hitting Apply
Comment 7 Peter Kümmel 2011-06-16 07:31:07 UTC
Looks like Kst interpolates from (0,0.225) to (1,0.809) which is in log scale
a line from (-350, 0.225) to (0, 0.809), so this is by design.
Comment 8 Nicolas Brisset 2011-06-28 20:39:52 UTC
OK, so what is missing so that we can squash this bug?
I seem to be able to reproduce it very easily, but Barth not?!??
Peter, do you also see it? If there are points on the concerned axis for which the value is <=0, kst should handle it gracefully, shouldn't it?
Comment 9 Netterfield 2011-06-30 02:53:32 UTC
SVN commit 1238831 by netterfield:

BUG: 272580
Fix the illusive Log mode range bug in psd's :-)
Directly edit parameters in plugin dialogs.
Fix plot selection in dialogs.



 M  +3 -2      libkst/vector.cpp  
 M  +1 -1      libkstmath/psd.cpp  
 M  +6 -3      widgets/curveplacement.cpp  
 M  +18 -3     widgets/scalarselector.cpp  


WebSVN link: http://websvn.kde.org/?view=rev&revision=1238831