Bug reports

No worries Maksym.

David

Hi David,

The bug: VStar opens a file even a user presses [Cancel] in Open File dialog: there is no difference between OK and Cancel. 

I could be wrong, but look in PluginComponentFactory.java, line 114:

[113]        // Was a file chosen or a URL string accepted?
[114]        boolean approved = true;

The code beneath the line only sets approved to true, never to false. Probably, 'approved ' should be set to false while initialization?

Is it a source of the bug?

(see also a picture attached)
 

Hi David,

I've tried to build the last revision (using ant) and have got errors connected with ExpressionVisitor.java (see attachment). It seems that VeLaBaseVisitor.java is missing. Could you please help to solve the problem?

Best regards,

Max

Hi David,

It's about a bug I've mentioned in #46  ("java.lang.IllegalArgumentException: null" when one uses AoV after DFT and vice versa)

Here is a scenario:

1) Start VStar
2) Open TEST_ASAS-SN_V.txt (see attachment).
3) Analysis -> DC DFT Standard scan..., do the scan. (*)
4) Close plot pane.
4) Analysis -> AoV with period range..., Min=1, Max=1000, Res=0.1, Bins=10
5) Click any of Top Hits (in the plot or in the table).
6) You will get an exception.

and vice versa:

1) Start VStar
2) Open TEST_ASAS-SN_V.txt (see attachment).
3) Analysis -> AoV with period range..., Min=1, Max=1000, Res=0.1, Bins=10. Do the scan. (*)
4) Close plot pane.
5) Analysis -> DC DFT Standard scan..., do the scan.
6) Click a Top Hit (in the plot or in the table).
7) You will get an exception.

See also a picture attached.

Quite annoying error. It requires the program to restart. I have not investigated it yet. Probably it is worth to create a ticket?

Regards,

Max

Thanks Max!

David

Hi David,

It seems I figured out why my SVN checkout did not get all files from VStar SVN repository. There is a "VStar.app" folder with "Icon " file. It seems that "Icon " filename has a trailing space (or spaces). Probably it has a special meaning under Mac OS, unfortunately, in Windows filesystem (NTFS) such a name is causing problems (formally, it is possible to create a file having trailing spaces in its name however most programs get confused with it). This is why (probably) I did not see "build.xml" file along with some other documents (however, despite the error, all *.java code were imported in my local folder so I could build the program!). Previously I ignored this error however now I suspect that it is probably a cause of incomplete repository importing.

I also noted that you have commented out line 1164 in Mediator.java: 

// convertObsToHJD(starInfo);

so now VStar does not convert data to HJD automatically in case of mixed time scales. SVN message states that it is connected with "#595 Standalone HJD conversion plugin". Does it mean that in the future versions, a rule of conversion will be changed? 

Regarding DoubleField.getValue(): returning null in case of any error requires a caller to check for null values, there are places in the code where a caller does not check. It is not a very serious bug, anyway, a user will get a null exception. Yet I'll try to find those places (I hope...). IMHO, if DoubleField.getValue() does not catch VeLa exceptions, it looks somewhat inconsistent (IntegerField, for example, does not use VeLa interpreter and never throws VeLa exceptions, it always returns null on error)

Regards,

Max

P.S. Another option would be using VeLa in both cases and always throwing an exception if a value is invalid or is out of bounds. It could be even safer.

P.P.S. Sorry for tediousness. DoubleField.getValue() uses VeLa interpreter, IntegerField.detValue() does not. So you can do some tricks while input in DoubleFiled (calculations on-the-fly)  yet can't with IntegerField (see the second picture: value for "Bins" will not be evaluated while values for period and resolution will be. Well, it is a feature, not a bug :) Another feature is that VeLa will evaluate 1/3 as 0 (integer division) and will not accept "1/3." only "1/3.0"... Well, I went too far, never mind :)

Hi David,

Could you check if the following bug is reproducible:

If uncertainty (mag err) in AAVSO Upload Format File = 0.0 and there is a record with the same magnitude in the input file, after loading the uncertainty of the record with zero error gets a value of the uncertainty of another record with the same magnitude.

An example data file is in the attachment plus screenshot.

Best regards,

Max

P.S. It seems I found a source of the error and a workaround.

I will describe it later, now I’m in a place with almost no Internet.

I would create the ticket but on Tuesday. Now I’m in a rural area with very slow Internet and a smartphone as the only option to connect.

Hi David,

I noticed that Vstar does not close files after reading (it prevents renaming/deleting them).
If another file is opened by the same plugin (or internal file reader) the first one is released however if multiple plugins are used, several files could be locked.
Many plugins use BufferedReader (inside retrieveObservations()) which never gets closed after reading. It could be enclosed in try-with-resource block (as recommended by JDK documentation):

try (BufferedReader reader = new BufferedReader(
  new InputStreamReader(getInputStreams().get(0)))) {
//... reading the input ...
} catch (IOException e) {
throw new ObservationReadError("IO error while opening input stream: " +
  e.getLocalizedMessage());
}

This closes the input stream immediately after reading.

A solution is slightly more tricky if a plugin creates BufferedReader in a separate method yet it could be solved too.

Internal file reader requires a separate approach.

I've made some tests with a plugin I’m developing now, it seems this approach works.

Don’t you mind if I create a ticket?

Regards,
Max
 

Hi David,

In NewStarFromObSourcePluginTask.java (Rev. 1642):
  in createObservationArtefacts() method (where retrieveObservations() is called):
why MessageBox.showErrorDialog() [line 297] is commented out? 
VStars swallows exceptions thrown inside plugin's retrieveObservations() (IO errors, for example) without any notification to a user which could be confusing.

Regards,
Max

Hi,

I'm Carlos from Argentina and I'm just learning about light curves analysis.

My operative system is Debian GNU/Linux with icedtea 1.8.3 and openjdk 11.0.4.

I have downloaded Vstar from https://www.aavso.org/vstar. When I initializes the program with 

javaws vstar.jnlp

 from terminal it works fina, but when I try to load a Star Data from AAVSO database I obtain an error:

javax.xml.parsers.FactoryConfigurationError: Provider for class javax.xml.parsers.DocumentBuilderFactory cannot be created

The same problem is running from Firefox.

However, with the same OS and java version, I downloaded VStar from SourceForge, run with

./VStar.sh

and now is runnig fine.

Do you have idea about why I can't run it from the web?

Best regards,

Carlos

Hello.

I tried many stars, but only some of them showed with the plugin. Mostly of them doesn't show light curve. Is this ok?

As part of the processing pipeline, Gaia identifies sources that appear to be variable using a set of machine learning algorithms. As of the second release, more than 0.5 million variable sources have been identified. This is not every variable you might find in VSX.  This set of variables is expected to expand with the third release.  Only stars identified by GAIA as variable will have a GAIA lightcurve.

To determine if the star you are interested in is a GAIA variable, follow the steps outlined in the "Obtaining Gaiasource_id " section of the documentation.

I think the whole astronomical community is very excited about GAIA 3.

best regards,

Cliff

Hi david

I use the B-V quite often but after the latest upgrade I am getting numerous error messsages that seem to indicate I cannot read back to the master files at AAVSO.  I have updated all of the plugins, rebooted the computer but still get the same messages.

Any ideas?