Tue, 09/16/2014 - 14:09
If you have a VStar bug to report, please feel free to post here.
You may want to check:
- that no-one has posted the same report here;
- the GitHub bug tracker to see whether the bug has already been captured. Ultimately, this is where bug reports posted here will migrate to.
It appears the VStar only issues the "error in observation source" message in regards to the first line of data in a file. Once the first line has been accepted, any subsequent lines with bad format are silently discarded.
I ran into this because I must massage my input data files, and I was not properly handling the changes for one class of observation. So long as a different class was the first record in the file, I got no error for the broken ones and they disappeared.
Which format type is that?
Is this via File -> New Star from File... or via an observation source plug-in?
This is more a question than a bug. I often want to look at the uploaded data for my run from the past evening. When one opens the New Star icon/box, the current date box shows the current date (e.g., 2016 Jan 03) but the default JD value is xx.5. My uploaded images from the past evening may start at xx.6. Therefore, they are not shown on the light curve. At least that is what I see each time I do this?
It is easy to fix by adding one day to the current JD such as changing xxx90.5 to xxx91.5 and I do that, but is it possible to increase/change the default JD (e.g., to either xx0.99 or xx1.5)? In other words, make the default current JD very slightly in the future so all recent data including last night's data are shown?
So, just a question? Thanks.
Yes, I've often had
Yes, I've often had the same thought when loading data in VStar that I've just submitted to AID.
I've captured this in a ticket:
Perhaps it would be enough to add 0.5 days? Probably should just add 1 day.
The actual link to the ticket is: https://sourceforge.net/p/vstar/bugs-and-features/509/
Has anyone seen the preferences registry key error shown in the attached image under Windows 7 (or Windows in general)?
I think that one workaround will be to run the VStar application as administrator. Another is to change permissions with RegEdit (neither is ideal).
I've only had this reported by one person but that doesn't mean others have not had the same problem.
I've seen a related warning under Windows but not an error that stops VStar from working at all.
Note that the error shows WebStart but the user in question reports the same error with the executable from SourceForge.
Cannot reproduce under Win7, Win10 (non-administrative account...)
Oh, sorry... This was an error from 2016!
Well, it's 2021 now. Hm...
No worries Max. I've seen a few come through from awhile back. Also, I notice that now when an email notification is sent re: a forum post, there is no link to the exact topic reply, just the topic, so you have to go to the last post and work backwards.
I posted a question about additive load in this forum yesterday in reference to TY UMa. I am currently enrolled in the VStar CHOICE course. Today I worked through the exercise described here for that course and could not get it to work. JoDee suggested I make this known here.
I have attached the ASAS file I loaded for R Car in step 8 of the exercise. I have attached a screen shot of the AID load dialog from step 9. I end up with only the Visual band. The ASAS filter 1, 2, 3 bands are gone. Info shows 2178 observations instead of the 1168+2178 I expect. There is no dialog popup that I assume would inform me of JD to HJD conversion.
I am running the latest version of VStar from webstart via IE. IE is setup with Java 8 update 91. I have made sure all my VStar plug-ins are current and followed that with a VStar restart. The only other thing I can add about my configuration is that I initially had trouble with webstart and downloaded an EXE to run Vstar. This was ver 2.16.10. I offer this info in case the older version may have stored some state on my machine that conflicts with the webstart version.
I don't know if other plug-ins could be causing a problem. If you would like me to remove all except ASAS and try again, just let me know.
the link in your post to the VStar exercise doesn't work for anyone who isn't registered for that course so I can't see exactly what the exercise is asking. However, I think there might be a little confusion about what the ASAS data file actually contains.
All observations in the ASAS-3 database are through a V filter. There is a separate database with I filter observations but your RCar-ASAS.txt file only has V data.
The ASAS telescope imaged a series of specific fields each night with some overlap between adjacent fields so a particular star will usually be in 3 or 4 different fields. In the RCar-ASAS.txt file you will see "#dataset= 1" which has all observations from one of the specific fields. "#dataset= 2" contains all observations from a second specific field, and "#dataset= 3" contains all observations from the final specific field containing R Car.
If you load the RCar-ASAS.txt file in VStar usingFile/New Star From ASAS File you should get the following plot:
Red data symbols are dataset= 1 and are labled ASAS-1 in the light Curve Control panel (View/Plot Control). Green data symbols are dataset= 2 and are labled ASAS-2, blue data symbols are dataset= 3 and are labled ASAS-3.
The VStar ASAS plugin rejects any observations graded C or D so the number displayed will be less than the total number of observations in the file.
I hope this helps. Cheers,
Thanks for the clarification on ASAS. I did not understand that.
See also David's update on additive load.
Thanks for providing additional info to Cliff about ASAS.
The fundamental problem Cliff and others have been seeing the last couple of days is that if you load some dataset (ASAS or any other), then load another via File -> New Star from AAVSO Database with a request for additive load, the additive part of the load will fail.
This is a regression introduced in 2.17.0. There'll be another release in the near future I suspect.
This has just been fixed in VStar release 2.17.1. Thanks to Cliff et. al. for finding this!
This is a request to add a feature to Vstar that would be fery useful. Quite often I upload a data set, but want to analyze only a portion of it. Vstar is very handy to display a subset of a data set. It would be great to be able to select a subset of a data set from the light curve display, and then do for instance a phase curve, or other analysis.
This can be accomplished manually, but one must note the JD, reload the data, and analyze. This feature would facilitate analysis greatly.
This may go some way toward helping. See page 43 of the use manual. Recenty added (a couple of versions ago) is the ability to create a: "...filter from the current plot view (View → Filter From Plot)".
So, you can zoom in on just the part of the plot you want visually, create a filter as above, then analyse that.
Does that make sense?
A variation on that would be to just be an option in the "which series?" dialog of DCDFT etc to just use the visible JD range. The filter from plot menu item does this, just at the cost of creating a filter.
I will try that. It should work for me.
I also have seen the extra zero bug in AM Her and DQ Her, FWIW.
I am doing some phasing for QQ Vul and when I pull up the default Epoch for phasing, its listed as "24048446.471" with an extra zero. It should read 2448446.471. VSX lists it ok. Not sure how this could happen. It does change the phase by 0.1 on the things I am looking at. The period at .1545204 is exactly correct.
Yes, I'm sorry about that. You are seeing a bug I just found a day or so ago:
On the surface, it appears that the Info dialog is showing the epoch in exponential format (with E7 at the end), yet the bogus zero doesn't quite seem to fit with that when expanded.
Are you seeing the epoch value for the first time in the phase plot dialog?
Sara and I were last night talking about another release in the not too distant future.
I had a complete hard drive failure and just reloaded VStar (the jnlp version). When I click on the plug-in manager to reload the plug-ins I get a message box that says Error reading remote plug-in information with an OK and a cancel button below the message. When I click either OK or Cancel I get another window that has a title in the top border
Plug-in Manager: Initializing Plug-in M ... with an OK and a canel button in the window. It doesn't seem to matter which button I click. Nothing happens.
There does indeed seem to be a problem with the 2.17.1 plug-ins access server side. We'll get onto that. This won't require a new VStar release, but a server-side adjustment.
You can workaround this for now in two ways:
Apologes for the problem.
I suspect that one or more of the current CHOICE course participants may have been seeing this since the 2.17.1 release.
That was my fault and it is now fixed.
Sorry for the inconvenience.
With the problem fixed, you can click the Set defaults button shown in the screenshot in my post above.
That will change the URL back to the 2.17.1 location.
Running VStar under OSX, it appears that a ':' character in a file name is displayed as a '/' in the load-from-file dialog. Version 2.17.1.
Attached is an example. Filename ends with "06:49.csv" but is displayed as 06/49.csv
You're right, Tom. Short answer:
«Once upon a time, before Mac OS X, : was the directory separator instead of /. Apparently OS X 10.7 is still trying to fix up programs like that.»
Boring tech stuff:
Wow. I was unaware of that aspect of OS 9 / OS X history.
So Tom, what is the effect? Does the file still load or does it have to be renamed?
The period search tool under "Analysis" is no longer working when a period range to search (as opposed to a frequency range) is the input. It goen into some loop that never ends. This was working about a month ago, but I see there has been a change in the program. The frequency ranve searches seem to work, but these are less convenient for me as I (actually my students) want to specify the resolution of the search in days.
Are you using the latest version (2.18.0)?
It will be best if we work with a specific example so I can try to reproduce the problem.
Can you tell me what data you are working with including JD range, and what parameters (low and high period, resolution) you enter into the DCDFT with period range parameter dialog box?
Finally had time to get back to the discussion. We are working on several data sets for the red variables in the globular cluster M13. It would be easiest if I could send you a sample text file that we have successfully run in the past (on two independent computers with the same results) but which do not work now. I just tried running one data again (header says version is 2.18.0), and it ran 5 minutes with no result using 1/10th the resolution we had been using that took less than a minute to do the search.
I uploaded the file below. Note, I leave for a three week trip overseas this Friday and may not get to this until I get back. In the mean time, we usually search 20.0 d to 200.0 d at 0.01 d resolution. Here are the first two and last two lines of one data file for the JD range:
2453932.6649 12.056 0.035
2453934.6281 11.999 0.038
2455760.6185 11.896 0.054
2455763.6711 11.872 0.041
Here's the result of running DCDFT with period range on the text file you supplied with min period = 20, max period = 200, resolution = 0.01. It completed within a few seconds on my 5 year old Mac.
I'm wondering whether the result window with periodogram is hidden, even though it should not be.
Is this on Windows? Which version? WebStart downloaded or downloaded from SourceForge as as zip?
This was awhile ago. Remind me: did we resolve this?
I get the starting application /verifying application flag and it just hangs there, sometimes it works, sometimes it hangs. Windows 7 pro, fast internet, java installed and working..
I seem to have missed this. Is this still happening?
So, I have my new super duper ASUS computer with scads of memory and the latest and greatest NVIDIA GeForce 1080 graphics card with 8 GB of ram on the card and 64 GB of computer ram memory. When I try to run VStar it loads but almost imediately after I click on any menus or control buttons in VStar the screen starts jumping around "fracturing" and partially duplicating itself on top of itself I am running windows 10 version 1703. I have an old HP computer running windows 10 version 1607 on which VStar runs perfectly and it also runs perfectly on an Alienware running Windows 7.
The NVIDIA driver is 18.104.22.16878 dated 4/12/2017. So the driver is current and the online driver update says I have the latest and best driver for the graphics card. Mouse and touch pad have hover and tap disabled.
I ran Windows Auto-repair and no problems were found.Anyone have any ideas how to solve the problem.
I looked for a ticket in SourceForge about this but did not find one.
VStar 2.20.0 Stand-Alone or WebStart
Windows 10 latest version
64 bit computer with Java 1.8 64 bitr and 32 bit intalled
Data: Del Cep Visual 2450330 - 2450380 about 8 complete cycles.
Load data and run Fourier plug in 1 period at 5.366266d and 1 harmonic
Switch to phase plot, turn off all series, set analysis to resids, set phase bins to 0.1 click apply and turn on model resids and means series plots in the visibility section.
display info dialog
Look at screenshot WBY_DelCep_1ch1_Phase PlotResids_0.10ph_Means.png
Close info dialog
Change bin size to 0.2 and click apply and redisplay info
Look at screenshot WBY_DelCep_1ch1_Phase PlotResids_0.20ph_Means.png
Note that the degrees of freedom are the same as are the fstatistic and p-value. Hmmmm?????
Repeat with bin size set to 0.3 and triple check that apply was clicked. by closing and reopening plot control twice to check that 0.3 was set.
Look at screenshot WBY_DelCep_1ch1_Phase PlotResids_0.30ph_Means.png
All the statistics remain the same! These three ANOVA should have different degrees of freedom since the number of bins are different. The f statistics and p values should not be identical. The displayed results are what you would expect from the f-test for regression. However, you don't apply that test to the residuals. Also I think the degrees of freedom would be 2 and 334 - (K-1) & (N-K),where N is the number of points and K is the number of coefficients, B0, B1 and B2 (the constant, sin and cos coefficients) for a single frequency fit.
Reload the data from tha AAVSO database
Run Fourier model as before. with raw data plot displayed. select residuals as the source of analysis set bin size to 1 day, click apply, turn of all plots except means and residuals and open the info box
You get the display as in WBY_DelCep_1ch1_Resids_1D_Means.png
close the info box
Now switch to phase plot change the bin to 0.1 phas steps leaving resicuals as the as the source for analysis and the display as resicuals and means
open the info box
which displays the info as in
WBY_DelCep_1ch1_Phase PlotResids after LC ANOVA_0.10ph_Means.png
The info dialog is still showing the statistics for the ANOVA made to the residuals of the light curve. The ANOVA performed on the phase plot only has 10 bins. so the first degree of freedom should be 9 and the second should be 327.
Sorry I think this calculation is broken.
ANOVA as displayed in the Info dialog only reflects raw mode changes.
The user manual is not helpful re: this currently.
What you need is the Current Mode ANOVA plugin:
This ticket is relevant:
I've just increased the priority of this.
So there's a solution, but it's not obvious.
Aha! and that is probably why it is Called Current Mode ANOVA. It works in whatever your current plot mode is. It works like a charm by the way. See Attached
You are a gentleman and a scholar and on top of that there is a rumor that you are a pretty good programmer as well.
Thank you Brad.
Glad the plugin is working well for you.
1. In the "online" version of Vstar, both on OS X and on Windows, the program preferences are not saved. In the "offline" version this problem is not present.
2. Both on OS X and Windows after a while it is impossible to mark a data discrepancy on the chart. The error message is java.lang.IndexOutOfBoundsExceptions: Index: [SomeNumber] Size: [SomeNumber]
3. When you exit VStar, at least on Windows, a "Java (TM) Platform SE binary (32 bit)" process remains active, which keeps the memory blocked and the opened files locked, for example by renaming or moving them in a different folder.
I have a new laptop with 3840x2160. The icons in the bar below the menu are not being scaled; they are really tiny.
Perhaps there is some help at this link.
Try loading alf ori from 2458360 to 2458460 and fit a quadratic polynomial to either B or V. You will get an IndexOutOfBoundsException. Cubic also fails, but linear and quartic succeed
VStar 2.19.0 and 2.20.0
With 2.20.0 I can't reproduce this currently.
I just loaded alf Ori with that JD range and created polynomial models of degree 1 to 6 without an error.
Can you repeat this and send me the log via File -> Log... please?
Can you also tell me what version of VStar you're using and what version of Java (java -version in a command prompt / Terminal)?
VStar 2.19.0 and 2.20.0
java version 1.8.0_181 and _191 running on ubuntu 16.04.10
Dec 07, 2018 4:34:43 PM org.aavso.tools.vstar.ui.dialog.MessageBox showErrorDialog
SEVERE: Polynomial Fit Error
java.lang.IndexOutOfBoundsException: Index: 11, Size: 11
at org.aavso.tools.vstar.util.ApacheCommonsDerivativeBasedExtremaFinder.find(Unknown Source)
at org.aavso.tools.vstar.util.AbstractExtremaFinder.toExtremumString(Unknown Source)
at org.aavso.tools.vstar.util.AbstractExtremaFinder.toMaximaString(Unknown Source)
at org.aavso.tools.vstar.util.AbstractExtremaFinder.toString(Unknown Source)
at org.aavso.tools.vstar.plugin.model.impl.ApacheCommonsPolynomialFitCreatorPlugin$PolynomialFitCreator$1.execute(Unknown Source)
at org.aavso.tools.vstar.ui.task.ModellingTask.doInBackground(Unknown Source)
at org.aavso.tools.vstar.ui.task.ModellingTask.doInBackground(Unknown Source)
It's failing during extrema determination.
OpenJDK or Oracle? Should not matter for the core classes involved.
I'm not sure why this would be different between our Java environments. Are you specifying the Visual series for polynomal fit?
What's the state of things after the failure? VStar is still running but no model exists?
In any case, I think I have found the bug and have created a ticket for this:
Also happens under OS/X 10.12.6
Fitting B or V data. VStar keeps running, but no model is created. I'm not sure how to determine the origin of the installed JAVA.
I've fixed this bug and it will be available in the next release.
I've been intensively using 'AoV' plugin for period determination. However, after running this plugin (which itself works well) VStar demonstrates strange behavior: when I run Period analysis (DC DFT) and then click on maximum I constantly get an error (see picture). The same error occurs if I try to select an item from "Top Hits" list. Only restarting VStar helps. This makes impossible to build a model etc.
P.S. I'm using VStar under Windows 10 (under Windows 7 the result is the same)
Sorry for the delay.
To make it easier for me to repeat this error, can you provide a scenario, i.e. target, JD range or just a data file, AoV params used, DCDFT params used, essentially the steps you followed to get this error.
Strange, now I cannot find a way to reproduce the behavior. Sometimes I get the error, sometimes not.
I'll try to make a scenario which always leads to the error and will inform you.