jasp-issues icon indicating copy to clipboard operation
jasp-issues copied to clipboard

[Bug]: ICC calculation in the reliability module does showinh only "." for output values.

Open utkb opened this issue 11 months ago • 24 comments

JASP Version

0.18.3

Commit ID

No response

JASP Module

Reliability

What analysis are you seeing the problem on?

ICC

What OS are you seeing the problem on?

Linux, Flatpak, Ubuntu

Bug Description

ICC calculation in the reliability module does showinh only "." for output values. For double checking, i used "Interrater Data from Shrout and Fleiss (1979)" from JASP data respiratory.

This is normal output: resim

But when i click refresh: resim

The output values disappear.

Expected Behaviour

I expect the normal output with values.

Steps to Reproduce

  1. Import Interrater Data from Shrout and Fleiss (1979) or any appopriate data.
  2. Refresh all calculations.

Log (if any)

JASP 2024-03-07 11_17_07 Desktop.log JASP 2024-03-07 11_17_07 Engine 0.log

Final Checklist

  • [X] I have included a screenshot showcasing the issue, if possible.
  • [X] I have included a JASP file (zipped) or data file that causes the crash/bug, if applicable.
  • [X] I have accurately described the bug, and steps to reproduce it.

utkb avatar Mar 07 '24 08:03 utkb

@utkb thanks for creating the issue. I cannot replicate this on my MacBook. Maybe someone with a linux can... @tomtomme ?

juliuspfadt avatar Mar 07 '24 10:03 juliuspfadt

Thank you for your answer and effort. I checked this issue on a fresh install of Ubuntu 22.04 version, and i can confirm this issue effects also fresh install. But i checked if problem is dependent on language settings. If i change system language and format to English, there is no issue.

utkb avatar Mar 07 '24 12:03 utkb

@utkb Your screenshot shows the problem using the english locale. But now you say with english it is fine? How is that possible? What locale did you test besides en?

And what about a fresh ubuntu install + updates available for 22.04?

On Manjaro I can also not reproduce (neither with 0.18.3 nor 0.19beta, english, german and spanish)

tomtomme avatar Mar 07 '24 12:03 tomtomme

@utkb Your screenshot shows the problem using the english locale. But now you say with english it is fine? How is that possible? What locale did you test besides en?

And what about a fresh ubuntu install + updates available for 22.04?

On Manjaro I can also not reproduce (neither with 0.18.3 nor 0.19beta, english, german and spanish)

Yes, it is strange but I was not changed the JASP locale settings but system locale settings.

utkb avatar Mar 07 '24 13:03 utkb

@utkb thx. that answers my first question. I cannot replicate this on manjaro however. I use german as system locale, switched to english and back to german via plasma settings and via manjaro settings manager and refreshed the analysis at each step in jasp 0.18.3.

Remaining questions:

  • [ ] What locale did you test besides en?
  • [ ] What about a fresh ubuntu install + updates available for 22.04?
  • [ ] And if you have time to test: What about ubuntu 22.10 or even 23.10 etc.? Maybe it is a flatpak bug that is solved already, but it may not be shipped with ubuntu 22.04 (yet).

tomtomme avatar Mar 07 '24 13:03 tomtomme

I realized issue in Turkish locale. I will test for last two question ASAP.

utkb avatar Mar 07 '24 14:03 utkb

What locale did you test besides en?

The problem is in the Turkish Locale.

What about a fresh ubuntu install + updates available for 22.04?

This not solved the problem.

And if you have time to test: What about ubuntu 22.10 or even 23.10 etc.? Maybe it is a flatpak bug that is solved already, but it may not be shipped with ubuntu 22.04 (yet).

I tested it with fresh install of up-to-date Manjaro XFCE. Selected Manjaro because rolling relase and have new version of packages. This propblem exists in Manjaro too.

utkb avatar Mar 13 '24 06:03 utkb

Thx for the update.

@JorisGoosen @RensDofferhoff Any idea why turkish -system- locale does mess with JASPs reliability module with flatpak?

tomtomme avatar Mar 13 '24 07:03 tomtomme

Well, maybe its not an UTF-8 locale?

Perhaps @utkb can post the output from the command locale in the turkish locale with the problem.

JorisGoosen avatar Mar 13 '24 11:03 JorisGoosen

[utkub@utku-standardpcq35ich92009 Masaüstü]$ locale LANG=tr_TR.utf8 LC_CTYPE="tr_TR.utf8" LC_NUMERIC=tr_TR.UTF-8 LC_TIME=tr_TR.UTF-8 LC_COLLATE=tr_TR.UTF-8 LC_MONETARY=tr_TR.UTF-8 LC_MESSAGES=tr_TR.UTF-8 LC_PAPER=tr_TR.UTF-8 LC_NAME=tr_TR.UTF-8 LC_ADDRESS=tr_TR.UTF-8 LC_TELEPHONE=tr_TR.UTF-8 LC_MEASUREMENT=tr_TR.UTF-8 LC_IDENTIFICATION=tr_TR.UTF-8 LC_ALL=

utkb avatar Mar 13 '24 12:03 utkb

Thanks @utkb !

Is "utku-standardpcq35ich92009 Masaüstü" your hostname by any change ? As in, what is the output of hostname?

JorisGoosen avatar Mar 13 '24 13:03 JorisGoosen

@utkb Oh no! If its the infamous username / hostname bug again... you will need to shorten your host / username to make it work and maybe also remove all ü ä spaces dots etc. R cannot handle certain long hostnames. See duplicates https://github.com/jasp-stats/jasp-issues/issues/1131 https://github.com/jasp-stats/jasp-issues/issues/2289 https://github.com/jasp-stats/jasp-issues/issues/2598 https://github.com/jasp-stats/jasp-issues/issues/2561

tomtomme avatar Mar 13 '24 13:03 tomtomme

@utkb Oh no! If its the infamous username / hostname bug again... you will need to shorten your host / username to make it work and maybe also remove all ü ä spaces dots etc. R cannot handle certain long hostnames. See duplicates #1131 #2289 #2598 #2561

I take this hostname from Virtual Machine Manjaro System that i install for testing the issue. On my actual system my hostname "utku". But i have same problem here.

utkb avatar Mar 13 '24 14:03 utkb

So @utkb at your normal system its basically something like utku@utku right? No spaces there then...

@tomtomme could you try the same turkish locale? (Saves me setting up a linux system just now ;))

JorisGoosen avatar Mar 20 '24 08:03 JorisGoosen

testing it

RensDofferhoff avatar Mar 20 '24 08:03 RensDofferhoff

@JorisGoosen So it works on seemingly all locals except Turkish. So that explains why we never noticed. I have no real explanation aside that some dep 50 layers deep doesn't handle a conversion? Googling utf8 turkish shows a bunch of issues.

RensDofferhoff avatar Mar 20 '24 09:03 RensDofferhoff

Yes, we normally not alone, if a problem exists in Turkish, also can be found in French and others have special characters.

I can use normally with English locale, but I can run all tests for solution of problem. Thank you very much and please ask me if you require any additional tests.

utkb avatar Mar 20 '24 09:03 utkb

I have tested 10 locales including french but none show the same issue. Quite mysterious

RensDofferhoff avatar Mar 20 '24 09:03 RensDofferhoff

So weird... Especially given that there shouldnt be any conversions at all anyway if its all in utf-8, especially on linux...

JorisGoosen avatar Mar 20 '24 12:03 JorisGoosen

So this seems all very unfortunate but I guess we have to close this one now. Since this seems not to be fixable, right?

tomtomme avatar Mar 20 '24 19:03 tomtomme

So this seems all very unfortunate but I guess we have to close this one now. Since this seems not to be fixable, right?

well, it might be fixable. if it is reproducible on other linux systems then the r-maintainer could put a vm in turkish and narrow down which particular part of the analysis breaks. Sounds like the sort of thing that ought to be fixed to be hones

JorisGoosen avatar Mar 21 '24 14:03 JorisGoosen

@JorisGoosen Yes, but it could only be fixed by the r-maintainer? Or does someone from the JASP team maintain this package, @juliuspfadt ?

tomtomme avatar Apr 22 '24 10:04 tomtomme

Well, we dont know what exactly is broken. First step would have to be to narrow down the actual culprit here, only then do we even know who might or might not fix it...

JorisGoosen avatar Apr 23 '24 12:04 JorisGoosen

This issue will be automatically closed in 42 days due to inactivity. Feel free to leave a comment if you believe this is still relevant.

github-actions[bot] avatar Jul 07 '24 01:07 github-actions[bot]

Automatically closed due to inactivity.

github-actions[bot] avatar Aug 18 '24 01:08 github-actions[bot]