Dependency Cycle - trying to have a calculation for a question reference itself from a previous repeat given certain logic

[j-bromp]

Hi,

Background info: I am creating a survey for people to go out and test different materials at different locations. The user may need to conduct an unlimited number of tests so I have contained everything in a repeat. On the first repeat, i.e. position(..) = 1, the user is asked to input their location. The question is a text field and their response would be something like 'Road 1' or 'Stormwater Line'. Call this ${location_initial}

On the second repeat they are presented with a select_one question asking them if they have changed location. If they select 'Yes' then a new question appears asking them to input their new location, and if they select 'No' then the initial location is retained in a field in the background.

Calculation for background location field, called ${location_a}:

if ( position(..) = 1, ${location_initial}, if ( ${yesno_newlocation} = 'Yes', ${new_location}, ${location_previous}) )

The calculation for ${location_previous} is:

indexed-repeat(${location_a},${ndm_bs_repeater},${rp_num_bs}-1)

What these calculation's aim to do is check that if the user selects 'No' to being in a new location, then the location of the previous repeat is called in. It causes a dependency cycle though, I guess because I am referencing ${location_a} in the ${location_previous} calculation.

A short sighted way around this would be to change the calculation for ${location_previous} to:

indexed-repeat(${new_location},${ndm_bs_repeater},${rp_num_bs}-1)

The issue with this is that if a person tests at one location for the first repeat, then changes on the second but doesnt change for the third, fourth and fifth... the ${new_location} field doesn't have anthing to pull from after the third repeat because they kept answering 'No' to the 'Have you changed location' question.

Any help would be much appreciated! I have several fields to apply this logic to..... eek

[DougBrowning]

Yea it cannot be done this way.  123 works with Names as the way to track fields.  So it has no idea that FieldA in repeat 1 is different from FieldA repeat 2.  I have been working on this for 6 years now with no luck.  Lots of posts out there if you want to look around.

Only way I kinda got to work was javascript.  But for me my repeat was too big.

This post here has links to the two options that I have found. 

https://community.esri.com/t5/arcgis-survey123-ideas/quot-same-as-last-quot-option-for-repeat-record...

Hope that helps

[j-bromp]

Thanks Doug, I'll give your function a go. Out of curiosity how big was your repeat? mine is about 70 lines in the xls form and it wouldn't be uncommon for it to be repeated 100 times in a single response.

[j-bromp]

Hi Doug,

I've tried: 

begin group	testing	testing
begin repeat	ndm	ndm
text	location_initial	location initial
text	location_a	location a		if ( ${rp_num} = 1, ${location_initial}, if ( ${yesno_newloc} = 'Yes', ${new_loc}, pulldata("@javascript", "functions.js","sameAsLast",${ndm},${rp_num},${location_a})))
text	prev_loc	previous location
text	new_loc	new location
select_one yesno_newloc	yesno_newloc	yes no new location
integer	rp_num	repeat number		position(..)
end repeat
end group

as well as:

begin group	testing	testing
begin repeat	ndm	ndm
text	location_initial	location initial
text	location_a	location a		if ( ${rp_num} = 1, ${location_initial}, if ( ${yesno_newloc} = 'Yes', ${new_loc}, ${prev_loc}))
text	prev_loc	previous location		pulldata("@javascript", "functions.js","sameAsLast",${ndm},${rp_num},${location_a})
text	new_loc	new location
select_one yesno_newloc	yesno_newloc	yes no new location
integer	rp_num	repeat number		position(..)
end repeat
end group

But getting a dependency cycle for both :(, is this the correct implementation of the script?

Script i invoked:

function sameAsLast(ndm, rp_num, location_a) {

        if (position > 1){

                return ndm[rp_num - 2][location_a];

        }

}

[DougBrowning]

I think the way you have it loc_a has an if to prev_loc and that calls loc_a so that is the loop I think.

First test just having it show you the value without the calc in loc_a and see if that works.  Then go from there.

It has been years since I used the code.