View Issue Details

This issue affects 1 person(s).
 6
IDProjectCategoryView StatusLast Update
07824Bug reportsSurvey takingpublic2013-10-07 21:41
Reporterronny_todgers Assigned Toc_schmitz  
PrioritynormalSeveritypartial_block 
Status closedResolutionunable to reproduce 
Product Version2.00+ 
Summary07824: Group randomisation and group relevance - buggy interaction.
Description

I have a large survey which has multiple sections of randomised groups - within each section, some groups are relevant and should not be shown based on previous responses. the survey is presented in group by group format. The individual questions within each group do NOT have any additional relevance equations.

What happens is that occasionally a page is displayed showing just the header and navigation button - inspection of the dom shows that the group is on the page but the entire group is hidden with an inline display:none;
Pressing next fails because the questions are mandatory and so the participant cannot continue - leading to missing data.

I have not fully exampled the EM code - but looking at the dom - i see a possible cause:

<div id="group-34" style="display: none;">
<div class="group">
..
..
<!-- END THE GROUP -->
</div><!-- endgroup -->
</div>

And from the relevance fields:
<input type="hidden" id="relevanceG41" name="relevanceG41" value="1">

and the step:
<input type="hidden" name="thisstep" value="42" id="thisstep">

It seems that the on-page EM is is trying to control the visibility of a group which is not on the page ( because the group have been randomised ) - and consequently hides the only group on the page.

As a temporary workaround I'm inserting JS to reset the display property of the group - this doesn't seem to mess up the relevance as pages with irrelevant groups are never rendered to the client.

Steps To Reproduce

will upload a test case shortly if the information given above does not identify the bug

TagsNo tags attached.
Bug heat6
Complete LimeSurvey version number (& build)130423
I will donate to the project if issue is resolvedYes
Browserchome - ie8 - ie9
Database type & versionPostgres 8.3
Server OS (if known)debian
Webserver software & version (if known)apache 2.2.9
PHP Version5.2.6
Git branch / Pull Request URL

Users monitoring this issue

ronny_todgers

Activities

ronny_todgers

ronny_todgers

2013-05-10 16:55

reporter   ~25224

I am of course struggling to produce a replicable test case for this - however I have noticed that when everything is working correctly the relevenceGxx field matches the ID of the group div - And so my supposition about the cause of the group being hidden is broadly correct. - I can confirm that when this was happening the group div and relevenceG settign did NOT match even though there was only one group on the page

I have saved the progress of an attempt on the survey where this happened three times ( on different browsers - I saved the sate after each one and then tested a different browser ) - Is there some way of interrogating this saved state to further this investigation? We have 8000 people meant to be taking this test in a couple weeks and the idea of any of them being unable to continue leaves me rather cold!

ronny_todgers

ronny_todgers

2013-05-13 11:36

reporter   ~25231

I'd like to add that I will still donate if this gets fixed - the 48 hours wasn't really fair for a Friday night report.

There is clearly some gremlin at work such that the group chosen to be on the page does not match the group relevance inserted onto that page. The fact it is not clear at this stage how this has come about shouldn't effect the fact that code should not allow it to do so. Can someone confirm that the output indicated in my original report is not something that should be possible by design?

c_schmitz

c_schmitz

2013-09-19 13:42

administrator   ~26293

I can't reproduce this. I will need a reliable test case for this. Can you provide a small survey that does reproduce it?

c_schmitz

c_schmitz

2013-09-27 23:35

administrator   ~26410

Feedback please?

ronny_todgers

ronny_todgers

2013-09-30 12:00

reporter   ~26446

My apologies it was some months ago that I came across this - I had to work around it and released that instrument.

While I couldn't work out how to reproduce at the time I had a saved attempt on the instrument which consistently displayed the fault - is there some way of interrogating this saved state ( if it still exists in the database! )

Its not impossible that the bug is actually something specific to that saved state somehow because I struggled to reproduce the bug outside of that - in other words it could actually eb a buggy interaction which relies on saving and resuming progress. In any case if it would be possible to look at this saved state then that is the best I can do at this stage realistically - where woudl I find it?

c_schmitz

c_schmitz

2013-10-07 21:41

administrator   ~26605

I am closing this issue because nobody else reported this and it happened in a relatively early build of 2.0.

Issue History

Date Modified Username Field Change
2013-05-10 13:55 ronny_todgers New Issue
2013-05-10 15:16 ronny_todgers Issue Monitored: ronny_todgers
2013-05-10 16:55 ronny_todgers Note Added: 25224
2013-05-13 11:36 ronny_todgers Note Added: 25231
2013-09-19 13:42 c_schmitz Note Added: 26293
2013-09-19 13:42 c_schmitz Assigned To => c_schmitz
2013-09-19 13:42 c_schmitz Status new => feedback
2013-09-27 23:35 c_schmitz Note Added: 26410
2013-09-30 12:00 ronny_todgers Note Added: 26446
2013-09-30 12:00 ronny_todgers Status feedback => assigned
2013-10-07 21:41 c_schmitz Note Added: 26605
2013-10-07 21:41 c_schmitz Status assigned => closed
2013-10-07 21:41 c_schmitz Resolution open => unable to reproduce