Fix committed to master branch: http://bugs.limesurvey.org/plugin.php?page=Source/view&id=30000

@thedirtypanda That fix seems to only change labels, no logic? The commit seems unrelated

Its not fixed. Currently @p_teichmann is working on this issue.It was my fault, cause I took the wrong mantis id. Just wait.

Fixed in Release 4.2.7+200604

The bot closed this because of the wrong mantis id. Re-opening

Please find below what we have seen from this bug:

Influentiators

The following are not causes, but influentiators in the situation that makes things confusing
1 - Sometimes we get duplicate question_orders (on new questions)
2 - Sometimes, question_order is assigned from survey_id

Cause

The cause (from our point o view):
Ordering algorithm while fetching questions for left side bar (surveyadmin.php::567 - getAjaxQuestionGroupArray) is different than the one used for fetching questions for condition manager (conditionsaction.php::1309 - getQuestionRows).
As we are dealing with duplicate question_orders, that differen derived algorithm difference derived on different ordering results.
When user sees ordering on left side bar false expectations are created in what it should be show on the condition editor.

Our suggested solution?

• Reorder the questions as to refresh the question_order.
• Fix situations 1 and 2.
• Review this happening on v3 (code seems to be the same)

Why not applying a fix in the order criteria?

We thought about adding QID as second order criteria, as to have that when question_order is duplicate.
Still, there are already lots of DBs already with duplicates, so doing that may change unexpectedly the order.
So, fixing the order criteria could be good on a different LS version, when people are aware that some changes may happen.

As we are dealing with duplicate question_orders, that differen derived algorithm difference derived on different ordering results.

What does this mean? :D

Would removing the code duplication fix the issue?

Would removing the code duplication fix the issue?

Yes. A workaround would be reordering the questions.
Then the problem would be solved.

And of course not having duplicates will avoid the issue

I think that the deciding factor is how the order is done when the survey is executed.
That way of ordering should apply everywhere.

@gabrieljenik Would that be easy to fix?

Question editor is working fine now.
(There were some fixes about the order)

Still, some data may be still corrupted, as on this case.
We could add a check on the data integrity report? As for fixing, we could trigger some fix from there.
I would start by the data integrity check

As data integrity check about order was already there, we just reinforce the message :)
https://github.com/LimeSurvey/LimeSurvey/pull/1849/files

I am having a similar problem with a long survey in LS 5.3.4, not sure if I should submit a different bug report.

When going to the condition designer, not all the previous questions are shown under "Previous questions" tab, and the ones showing seem to be out of order. Even more, an existing condition for the same question, added a few months ago, shows "Variable not found: 225835X73112X963563" when listed inside the condition designer (screenshots attached).

Going to "Reorder questions & groups" and saving doesn't fix the problem. Checking the logic of the survey returns with no issues found.

Ginet

The problem I am seeing is mostly described in https://bugs.limesurvey.org/view.php?id=17968

Ginet

This bug is already fixed lastweek: https://bugs.limesurvey.org/view.php?id=17968

I am closing this bug report.