Hi all,
I'm running Vstar 2.18.0 under linux Ubuntu 16.04.
I just did try to reproduce and recreate the procedure used in Vstar Manual under the Period Analysis topicc. I did load the data of eta Aql from JD 2455695.8431 to 2456419.8604, as indicated on page 59 and performed a DCDFT standard scan on all Visual observations.
When I obtained the top hits, I clicked the "Find Harmonics" then "Create Model" using the 4 harmonics found ( page 65) and then "new phase plot".
When I select Analysis-->model-->select-->show model nothing happens, but if I create a polinomial fit everything seems to work properly ( the "Model information" window opens a and I can see model details).
Any suggestion?
Simone
SSIM
Hi Simone
Apologies for the delay.
I'm trying to replicate your result but cannot at the moment.
Do you see any errors in the shell from which VStar.sh is invoked?
When creating the model, did you select 4 harmonics from the drop down menu? That seems to be what you said you did.
I have attached a screenshot of what I see (also below) with 2.18.0 on Mac OS X. Perhaps that will provide some clues.
What version of Java is installed?
David
Hi David,
thanks for the reply
I see exactly what you see except for the last window ( Model Information) . It only works with polynomial fit..
I tried the same analysis on my laptop ( Win 7).... same result.
Here is the result when I look for java version:
simone@simone-EJ178AA-ABZ-t3240-it:~$ java -version
java version "1.8.0_111"
Java(TM) SE Runtime Environment (build 1.8.0_111-b14)
Java HotSpot(TM) Client VM (build 25.111-b14, mixed mode)
Seems to be ok...
I tried uninsitall and reinstall java....nothing.
The only error message I get is on Vstar launch : libpng warning: iCCP: known incorrect sRGB profile
Simone
Hi Simone
I'm trying to think what might be causing this.
Can others on the list reproduce this problem with 2.18.0 on Linux or any other OS?
David
Nothing....
I updated the OS to ubuntu 16.10, no result.
Then tried on ubuntu 15.04, both with Orcle Java JRE and Open JDK.... nothing
Simone
Well, I just tried Vstar in another computer, Windows XP Pro, Oracle Java 7.... extactly the same problem: show model button works only with polinomial fit and not with Fourier analysis. Mybe I should try to rebuild fronm source code on my computer...but I have no idea how to do it.
Simone
Hi Simone
I'll try to reproduce this on one of a Windows laptop I have access to. Anyone else want to bite? :)
This is odd because we know that feature has been working with 2.18.0 with various OSes, e.g. during the last VStar course, on my Mac recently etc.
You are welcome to rebuild it but it should not be necessary. See ReadMe.txt and extlib/ReadMe.txt that come with the full distribution on SouceForge. I use Ant to build.
David
Hi David and Simone
I have tried to replicate the error following Simone's instructions and I cannot replicate it. VStar is working as expected. I even tried opening VStar as a AAVSO member and as guest (just in c ase there was a difference) and it did not appear to make any difference.
Sorry I cannot be of more help.
Robert
Thanks for that Robert.
I wonder whether there's a step Simone may inadvertently be missing?
David
Well, I tried again.
Loaded data set for eta Aql from JD 2455695.8431 to 2456419.8604, then made a phase plot, a DCDFT standard scan and created a model with the top hit period and 4 harmonics, then a new phase plot. ...I still can't see the model data. My system is configured to use comma instead period as decimal separator, but I don't think it's important, because everything else works properly.
..... fortunately I still can use R for analysis...
Okay, that's useful information.
So far, only people with English locale have posted about not being able to reproduce what you see. Perhaps there's a localisation related bug. We've seen that before.
Can anyone on the list with a non-English locale reproduce this? Preferably Italian, but at least German will use commas instead of periods as decimal points. Others here have helped with locale related bugs and testing.
David
Ok, i tried changing language and numeric settings without success...
I also ran Vstar on a win XP machine, using english as default language and period as decimal separato...no success.
Also downloaded Vstar from SourceForge and ran in Win XP via the MS-DOS batch file. Here is a text file with the output I obtain on the console , hope it helps....I'm not a programmer , so I dont' know what it means. Le last part of the file is the one i see when I click the "show model" button.
Simone
Excellent, thanks Simone! This helps a great deal. I'll get back to you ASAP.
David
Hi Simone
I've created https://sourceforge.net/p/vstar/bugs-and-features/560/ to capture this as a high priority bug.
You weren't going crazy. This really is a bug.
Would you be willing to test a development version once I think I've fixed this?
David
ps - apologies if I seemed too skeptical
Hi Simone
An update on progress. I've implemented locale selection preferences, so that now you can change the locale at run-time via the Preferences dialog and without changing your operating system's preferences. Any locale change will be isolated to VStar.
So I set my locale to French, German and others to get the comma-delimited numbers behaviour and I have reproduced the error you saw. I'll now work on the fix and test it.
Having locale preferences has two benefits:
Stay tuned.
David
Hi David,
I will be glad to test the developement version !
Thanks for your help.
Simone
Thanks for your patience Simone.
David
I also ran Vstar under Linux Ubuntu 15.04.
This is the console output.
Simone
That output illustrates one of the undiminished beauties of pure Java: platform/OS indepdence in a way that few other languages/environments achieve.
Most importantly, it helps us with this problem.
David
Hi Simone
I've fixed the bug. I'll contact you re: details of a development release.
David
All good. The fix will be included in the next release.
David
Thanks Simone. The fix is in 2.19.0, whose release is imminent.
David