Skip to main content

Java error

23 posts / 0 new
Last post
Deconinck Michel
Deconinck Michel's picture
Java error

Hi

I'm new with VStar,

Everything seems ok while using one of the 10 example stars from the Citizen Sky group.

When trying to load another star today Khi Cyg in Other Star, I receive the error:

Java.lang.NullPoinerException: null

My environment is :

PC Dell with Windows 7 - MS version 6.1.7600 - VStar (in french ... why is that comming in french ???) 2.19.0 svn 1452 - firefox 55.0.3 (32 bits) - Java 1.8.0_144 - Java HS 64 bits - after loading VStar the free memory is 660 Mo

I have also an error related to the plugin : Observation source Plugin error, maybe not critical

I probably made an error somewhere but where ???

Can someone help me ?

 

Thanks for your answer

Michel

 

 

 

 

dedricksond
Same error note

I’ve been using VStar for quite some time now. I’ve been getting the same null point error message lately. Some stars are fine, others won’t load.

Duane

Deconinck Michel
Deconinck Michel's picture
Ok, thank you for your

Ok, thank you for your message Duane, I tried with SS Cyg as well with the same error... Two interesting stars for me today.

l will delete Vstar from my tool list.

Michel

dedricksond
VStar Error

I would never delete VStar from my tool list. It is a very powerful tool and usually works very well. I use it very often. I just don't know what is going on with a few stars. They are ones with a lot of data.

I've even used it in a lower level math class that I teach to demonstrate ways of looking at data and the usefulness of "simple" visual data where anyone can contribute. Between the last sentance and this one I loaded a DCEP and used the DCDFT to find the period using just visual data. The period found was within minutes of the published value. This took maybe 2 minutes and could be refined to get even better results.

Keep using it and think about the VStar Choice class for VStar. It is a great class.

Duane

dedricksond
VStar Error Again

Hello Again,

I just loaded khi cyg using dates other than the defualt and it worked fine. There is an older forum post around May 2016  that may have something to do with what is happenning. It concerns loading errors with ss cyg.

Maybe Ken or David could chime in?

Duane

WGR
WGR's picture
Try KHI CYG

I had trouble this morning using Khi Cygni.  I think KHI CYG is what worked with the LCG.  I suspect Vstar will only work with the same designation.--Well that did not work either.

Sorry

 

Gary

David Benn
David Benn's picture
Khi Cyg

Hi Gary

"Khi Cyg" works in LCG and I would expect it to work on VStar too. 

Can you confirm?

David

dedricksond
Me Again

Hello,

I'm able to load all data for khi cyg if I don't include the date JD2457910-11. There is one observation in V on that date. Maybe as in a previous forum, there is something funny going on with that particular observation report. I found this by looking at LCG for that date. (LCG loads fine, it sometimes takes time. If you choose fewer bands to load it is quicker.)

Duane

weo
weo's picture
Problem observation?

Thanks for the detective work, Duane. I'll investigate the data on khi Cyg and for that particular observation report and report what I find.

Elizabeth Waagen, HQ

David Benn
David Benn's picture
JD 2457910 to 2457911

Hi Duane, Elizabeth

Here is the web service URL for a lookup of Khi Cyg for that date range:

https://www.aavso.org/vsx/index.php?view=api.object&&ident=Khi+Cyg&data=100&fromjd=2457910&tojd=2457911&where=mtype%3D0+or+mtype+is+null

The second observation's uncertainty value is "NAN" (not a number).

Looks like LCG is being tolerant of that  by ignoring the observation but VStar is not. LCG just shows one observation, ignoring the second.

WebObs also shows the NAN:

https://www.aavso.org/apps/webobs/results/?end=2457911&obs_types=all&start=2457910&num_results=25&obscode=SGEA&star=Khi+Cyg

Two things should happen I guess:

  • VStar should be modified to handle that. I just remembered I already created a ticket for this (https://sourceforge.net/p/vstar/bugs-and-features/602/) and started addressing it not long before my Mac died (I have backups and committed the code remotely). I've added the Khi Cyg example.
  • The observation should be fixed via WebObs.

David

weo
weo's picture
Khi Cyg obs fixed

Hi David,

Thank you very much for determining the 'NAN' content of the uncertainty field for that observation. I have fixed it and the khi Cyg data now load smoothly for me in VStar.

Elizabeth

David Benn
David Benn's picture
Apologies for my lack of response

Hi Michel

Thanks Duane, Gary, Elizabeth for your help and also thanks for your kind words Duane. It's always great to hear about how people are using VStar in different ways.

I have limited access to compute resources right now. My Mac's charging port died a few weeks ago. It was several years old and due for replacement anyway. The replacement Mac is several days away (getting a bigger SSD than comes standard) .

While doing much debugging right now is hard I'll do what I can by looking at the web service response. If you have other star/date combinations for which you see errors, please post them here.

Michel, you can change the locale from French to English in Preferences. VStar is just picking up on what it thinks is the set locale for your opertating system. Does nothing else on your PC appear as French? Let us know here if you need help with this. See also the relevant section in the user manual.

Do you see the observation plugin error just when loading data from AAVSO or at other times too.

I'm sorry you are having a frustrating time. We are here to help.

David

dedricksond
Same Error with SS CYG

Hello,

I get the same nullpionter error with SS CYG. I can load all data except between JD2457972-73. I noticed one of the V observations are by the same observer as the single observation that seems to be responsible for the error with khi cyg. The observation date for ss is 2457972.66807.

I have no trouble loading either star in LCG other than the time to load lots of data.

Duane

tcalderw
tcalderw's picture
NAN

WebObs upload should reject NAN in *any* numeric data field - it is never a legitimate value.  VStar is certainly not at fault for not coping with this. 

weo
weo's picture
SS Cyg problem obs

Hi Duane,

I have been looking at that SS Cyg observation on  2457972.66807 (and all the SS Cyg observations for 2457972.6+) but I can't see anything wrong with any of the fields. However, something clearly is wrong as VStar will load up to 2457972.6 and from 2457973 on but not in between. It's not VStar, it's the data.

I'll keep checking, but unfortunately I have to leave now for the day to go to a meeting. I'm not sure I'm going to be able to work on it over the weekend, but if not, I'll pick it up Monday morning.

Elizabeth

lmk
lmk's picture
You can't blame the data if

You can't blame the data if LCG will load it! Clearly, something in VSTAR is causing the problem. Having done a lot of programming that interfaces with end users, you cannot be too thorough with ERROR HANDLING in the code. The "try"..."catch" constuct is a MUST in every line of code that reads user data. At the very least, even if the program hits some crazy input it cannot handle, each statement should have a unique error code dumped to the logfile, to allow tracing exactly where the problem originates from.

Mike

 

David Benn
David Benn's picture
Bug ticket

Hi Mike

VStar applies error handling and exception blocks liberally. Sometimes though, things "fall through the cracks"

As per comment #10 above, this bug has been acknowledged and capturred; see https://sourceforge.net/p/vstar/bugs-and-features/602

There are two things you can do with an erroneous observation: ignore it or try to correct it.

Not sure what LCG does without looking again. Perhaps someone else knows.

If I see invalid (not well-formed) input, my inclination is to ignore the observation but indicate that to the user. I'll admit that that policy is not applied uniformly across all observationn source plugins.

Having said all that, it's also been pointed out by at least one other person in this thread that as far as AAVSO submitted data is concerned, the best place to stop errors from making it into the database is WebObs. 

David

BSJ
BSJ's picture
error found in the database

Hi David and all,

I found the problem with the SS Cyg data. Some records by observer JM contained a value of "\2" for uncertaintly rather than the intended ".02" and my guess is that VStar didn't like the blackslash (since it is an escape character). I have corrected the observations and SS Cyg now loads correctly again.

I agree completely that this sort of thing should be caught by WebObs. Until we can get to that project, I think that probably the best thing for VStar to do is ignore such data issues. Unfortunately, the uncertainty field in our database is a varchar so MySQL itself will not stop strange characters from getting into that field.

I hope this helps,
Sara

lmk
lmk's picture
CHAR for numbers?

Seems odd that an uncertainty should be stored as a character type rather than a numeric/decimal/float type?

 

David Benn
David Benn's picture
Error and thanks

Hi Sara

Thanks for looking at all this. 

As it turns out, it's the string "NAN" being presented in the web service XML that VStar was having trouble with, which of course would have been caused by the \2 you found. See this comment above: https://www.aavso.org/comment/56296#comment-56296

As Michael said, it's interesting that the uncertainty field is a varchar but as you said, right now, that's the reality. I have had a fix committed for awhile that will be in the next release.

David

Deconinck Michel
Deconinck Michel's picture
Where is Khi Cyg ?

Hi

First of all, regarding the Java error, thank you for the correction work in the DB, I will try VStar again in a while.

But

I don't know if this is linked with the DB correction but it seems that the complete data set of Khi Cyg (or Chi Cyg) is lost (see attached file for the 75 last days, only the last 24h estimations exist)

Thank for your feedback.

Michel

File upload: 
BSJ
BSJ's picture
khi Cyg is back

Hi Michael,

The Khi Cyg data is back. We are doing some maintenence having to do with a VSX cleanup that temporarily kept you from finding that star's data. Nothing was lost and all should be OK now.

Many thanks,
Sara

Deconinck Michel
Deconinck Michel's picture
Ok everything looks ok now,

Ok everything looks ok now, super job Sara, Thank you +++

Michel

Log in to post comments
AAVSO 49 Bay State Rd. Cambridge, MA 02138 aavso@aavso.org 617-354-0484