Related to relevanceStatus : …

Partial fix (muts be improved)

• https://gitlab.com/SondagesPro/coreAndTools/reloadAnyResponse/-/commit/c7e59524e4398f940ba829de9b418ef4d965873f
• https://gitlab.com/SondagesPro/coreAndTools/reloadAnyResponse/-/commit/966794b13e3c50cb2822f43e25becaacf8b559d9

Then maybe a better solution are :

• create a maxpage column in database : have lastpage + maxpage
• when reload : LimeExpressionManager::JumpTo($maxpage, false, false, true); + LimeExpressionManager::JumpTo($lastpage, false, false, true);

@DenisChenu will you like to create a pull request with the partial fix? I will create a task for the better solution you have mentioned and it will be picked up at a later stage

By Partial fix : it's usage of the whole plugin … and clearly : this one is here for a lot of other things ;)

We have been reviewing this and trying to reproduce it.
We could :)

Here are some thoughts.

1) The response is completeley loaded from the database, that is, it is loaded with the data that had been previously completed.

2) When you advance to the first page (the one with the formula with the assignment) the expression is processed, which in this case is: DataSet = if (is_empty (DataSet.NAOK), 'prefilled', DataSet.NAOK)

DataSet.NAOK only returns the value it raised from the database if DataSet.relevanceStatus == true.

At that moment, relevanceStatus for DataSet is false, because that question has not been processed yet, so its relevanceStatus cannot be known.

When starting a survey and uploading the questions, it is very difficult to know the relevance status it had at the end of the previous upload since those are calculated on the fly. That makes .NAOK accessors to return always empty.

3) If the assignment is done on the same page it works, as relevanceStatus becomes true.

Suggested actions:

• Generate new attribute ".raw" to use on EM. No dependance on relevance status.

• If the assignment needs to use .NAOK (or similar), and that requires relevanceStatus to be already calculated, does the assignment really works for setting default values?

So far, I believe it does only if equation is set on the same page as in the question to be defaulted (and view mode is not question by question). So, either
a) I believe we should remove the "forward-assignment" part from the manual (set default value) or
b) Ammend the manual to reflect so.

a) I believe we should remove the "forward-assignment" part from the manual (set default value) or

? I really don't see how some existing survey can not work without …

Again : We already have a lot of issue with maxtstep !== real max step

Why not

• create a maxpage column in database : have lastpage + maxpage
• when reload : LimeExpressionManager::JumpTo($maxpage, false, false, true); + LimeExpressionManager::JumpTo($lastpage, false, false, true);

Why not

• create a maxpage column in database : have lastpage + maxpage
• when reload : LimeExpressionManager::JumpTo($maxpage, false, false, true); + LimeExpressionManager::JumpTo($lastpage, false, false, true);

I feel more comfortable adding .raw then updating the rendering process.
Still, I understand your point.

Just thkinking outloud, wouldn't it be just enough with:
LimeExpressionManager::JumpTo($lastpage, false, false, true); two times?
First one will be usefull for setting relevance.
Second one will be usefull for overwritting any values miss set on first pass.

LimeExpressionManager::JumpTo($lastpage, false, false, true); two times?

No the point is : when reloading you are never sure lastpage in response table are the max page.

Then you need to save :

• lastpage : the current page seen
• maxpage : the max page seen

And you need to move to the maxpage before move to lastpage

See :

1. https://gitlab.com/SondagesPro/coreAndTools/reloadAnyResponse/-/blob/master/reloadResponse.php#L69-75
2. https://gitlab.com/SondagesPro/coreAndTools/reloadAnyResponse/-/blob/master/reloadResponse.php#L164-167

It's really used like this and it work.
I try to use a response DB column : but it break

I stop here : i already have the solution, i already answer multiple time on the how to fix the issue.
If someone have this issue : he can contact me for an offer. I'm really tired on the way it work now.

And you need to move to the maxpage before move to lastpage
You are right.

Still, that's like a great update.

ping @c_schmitz @ollehar for comments.

Want to discuss on zoho but https://projects.limesurvey.org/portal/limesurvey#taskdetail/85781000000271009/85781000001007039/85781000001035097 seems deleted.

@DenisChenu it is not deleted. I added you to task owner and you should be able to comment now.

OK thanks :)
Seems zoho don't show the good error … … a 403 is really better.

I split this one in 2 part : one without updating DB

Fixing if user don't use previous , using lastpage only for master. Fix can be apply if OK in 3.X

Fixing if user use previous : adding an survey attribute and coiluln to keep current step (lastpage) and max step (maxpage)

Sh! i made an error …
Solution didn't work

set relevanceStatus to true by default work …
https://github.com/LimeSurvey/LimeSurvey/blob/16d809ea6cddb2fb14f9fbec3da12e17cf513569/application/helpers/expressions/em_manager_helper.php#L8771

@gabrieljenik : sorry, you're right … the only way is to fix the relevanceStatus point on a way or another …

Some idea :

1. set relevanceStatus to 1 by default
2. set relevanceStatus to 1 by default ONLY when come from NAOK or valueNAOK (or all)
3. add a new suffix (rawNAOK ?)
4. set the relevanceStatus of all questions before maxstep before do the jumpTo

2 can be great in my opinion ?
4 can be intereseting

When this is fixed : i fix the maxPage in develop version.

PS : for 2 : update the search here : https://github.com/LimeSurvey/LimeSurvey/blob/16d809ea6cddb2fb14f9fbec3da12e17cf513569/application/helpers/expressions/em_manager_helper.php#L8572-L8576

I need to get contexto into this again.
Feels like a big change, specially as is on the EM side.

@ollehar, what do you think?

Maybe add defaultRelevanceStaus as a public variable too ? Then we can set it when JumpTo after reloading.

I really like to have a clean way to fix it … really allow usage of "Using Assignment Operator" https://manual.limesurvey.org/Expression_Manager#Using_Assignment_Operator

A lot of forum helper (Jelo, holch …) give this advice : {QCODE=if(is_empty(QCODE.NAOK),"Value",QCODE.NAOK)} on forum …

I still need to get context.
Still, I am more inclined into adding new functionality as to no brake anything that is already working.

After a lot of thinking : I really think we need to set relevanceStatus of all variables before the current step

But relevanceStatus is set on ProcessAllNeededRelevance https://github.com/LimeSurvey/LimeSurvey/blob/538e0706f17938b31631c4b76874020323da7afd/application/helpers/expressions/em_manager_helper.php#L4102

But this function

1. return if $gseq != $this->currentGroupSeq
2. Seem not called when JumpTo with $force

Go to each group and https://github.com/LimeSurvey/LimeSurvey/blob/538e0706f17938b31631c4b76874020323da7afd/application/helpers/expressions/em_manager_helper.php#L6743 StartProcessingGroup ?

In a static function ReloadSurvey ?

@galads : not done ,but set as done in project : https://projects.limesurvey.org/portal/limesurvey#taskdetail/85781000001335039/85781000001337011/85781000001035097

https://github.com/LimeSurvey/LimeSurvey/pull/2211

Fix committed to master branch: http://bugs.limesurvey.org/plugin.php?page=Source/view&id=33165

Fixed in Release 5.2.14+220214