Hi

I've faced with the typical problem of Province > Town dependency. The additional problem is that I'm currently using an Array dual scale that is using a filter array based on the previous response.

I'm working with numerous categories so have been trying to simplify as best possible.

While my Province dropdown is neat enough, my Town dropdown is rather large (since I've had to put in all possible towns).

Searching the forum posts I landed on the autocomplete workaround here:
www.limesurvey.org/manual/Workarounds:_M...wers_for_text_inputs

I figured this may be better to use. I tried it in Array (Text) but no luck there. Is it possible to use this autocomplete on arrays?

(even though I'm working with two different variables I'm happy for them to have the same options if it will make my Town variable easier to work with).

Thanks!

You can use autocomplete on array, just need the good id.

If you are in question text
$('#answer{SGQ}XCODE_Ycode').autocomplete(states, { if i don't make error.

You can have a look to at gitorious.org/ls-selectfilter

Thanks Denis for the reply. I must admit, I'm not quite sure how to incorporate your suggestion?

Current I have the following for my question text:

Also, where would I use your ls-selectfilter?

Nice work Denis. We should add that to the workarounds.

Also, where would I use your ls-selectfilter?

You would place Denis' function in template.js or in the question source. Then you would apply the function to the question by placing this in the question source:
The answer codes for both scales must be the same length and the scale-2 options are filtered by the first n characters of the answer code, where n = (answer-code-length - 2). So, if you have 4-character answer codes, the filtering will be done by the first two characters. See screenshot below.

Attached is a working sample survey. In this survey I've also added a temporary fix for a bug in 2.05 that records data for unanswered dropdowns - bugs.limesurvey.org/view.php?id=8875 .

Wow, that is a very elegant solution. I'm going to go with it on my survey and report anything picked up. This certainly does improve things substantially!

Yeah, note that when the bug is fixed you can remove this part from the script in the question source of the sample survey:

Yes, I did pick this part up from your example survey - thanks. Will keep an eye out on the updates.

tpartner wrote: Nice work Denis. We should add that to the workarounds.

Thanks Tony,

It's one of 'future' question attribute in plugin. But here there are a problem with first system if question is hidden at start... i think i fix it somewhere but don't remind where ...

But here there are a problem with first system if question is hidden at start...

The only problem I can see is that the width of the cloned select is set to that of the original and if the row is hidden by filtering the original has no width.

Removing this line should do the trick:
Here's another survey with filtering applied to the array.

Yes, it's that, i think i fix it differently . And without this line the 'select' update his width (set to auto), i don't like it .

Denis

Denis, Tony - thank you for all your help!

This section now works like a charm! (finally tested)

If I have any nit picks it would be that the options "jump" when selected. So as soon as I select my first option, both options shift to the right.

This script is a fantastic solution! Expecially for italians users...
In Italy we have 110 Province and 8.057 Comuni (cities).
I've adapted the script in order to have a unique ID for cities in each province (3 chars instead 2), beacause in many cases we have more than 100 cities for province.
In this example the lenght of answer code for both scales is fixed at 5 characters (maximum for anser code). First two characters is abbreviation of province (RM is Province of Rome). Lasts 3 characters is the unique ID for both scales, expecially for the Cities, so the maximum number of cities for province now is 999.
I've modified the scritp in the "Substring" arguments. For example:
Old: substring(0, $(this).val().length - 2)
New: substring(0, $(this).val().length - 3)
I attach a sample of this solution, with the modified script (I also commented).
I ask to Denis and Tony the permission to publish the complete solution with the modified script in the italian forum.
Thanks a lot for this discussion/solution.
Luciano

The script don't have actually a licence ?

MIT surely.... (then yes ... you can copy /paste / adapt / etc ...

And yes : MIT at starting : gitorious.org/ls-selectfilter/ls-selectf...d:selectfilter.js#L3

I've added a comment to documentate my modify. This:
/*
Modified Version by Luciano Fanfoni 04/04/2014
Fixed Lenght of both Answer Code (scale 1 & 2): 5
Firsts 2 chars used to filter. Lasts 3 chars used to identify unically options of scale2
Example of Modified code:
Old: substring(0, $(this).val().length - 2)
New: substring(0, $(this).val().length - 3)


Distributed under MIT licence
*/

Hi Guys,

Back on this thread which I struggle a bit to understand. I have 2 questions here:

1. is there somewhere a summary of how this can be tackled (Denis maybe)? as there seem to be different versions, with bug fixes, etc. Any sort of a "final word" on this?

2. anything "more robust" planned along those lines in future releases?

Cheers!