Bug for a computed variable in a loop

[romaintailhurat]

Describe the bug

A calculated variable on a questionnaire is having a bad structure.

The computed variable T_MENAGE_PRINC from the TCM has in a problematic use case this form:

"T_MENAGE_PRINC": [
      1,
      0,
      [
        0,
        0
      ]
    ],

That is, instead of having n scalars in its array (here n = 3, because the loop have 3 iterations), we have the last index being an array of 2.

• [x] I tested it on the storybook, which leads me to believe that it's a lunatic bug.

To Reproduce

In order to reproduce, i attach here two questionnaires:

• the first one srcv-buggy.json has the bug,
• the other srcv-placebo-ok.json don't.

To reproduce the issue:

1. With srcv-buggy.json on the storybook 2.7
2. Choose a number, for example 3 for this question
3. In the following question add a surname for each inhabitants
4. You can skip answers until the birthdate that should make the individuals adult (use 2000 for example)
5. Skip everything until the pairwise question in which everyone is a housemate ("colocataire")
6. Skip to the "other house" question and answer "No" for everyone
7. At last, we arrived to the "test" Sequence which display various variables including T_MENAGE_PRINC and we can see the last one is not working (because of the structure highlighted above)

Source questionnaires

srcv-bug.zip

[Grafikart]

@romaintailhurat Quel est la différence entre les 2 json ? Que représente la variable T_MENAGE_PRINC ?

[romaintailhurat]

La variable T_MENAGE_PRINC est une variable de boucle sur chaque habitant du logement pour déterminer s'il fait partie du ménage principal, sa valeur est 1 ou 0. Le code VTL:

if ($T_TYPE_MEN_LOG$ = 1 or $T_TYPE_MEN_LOG$ = 2 or $T_TYPE_MEN_LOG$ = 3 or $T_TYPE_MEN_LOG$ = 5) then 1
else if ($T_TYPE_MEN_LOG$ = 4) then (
    if ($PRENOM$ = $PRENOMREF$) then 1
    else 0)
else (
    if $T_NB_MEN_FINAL$ = 1 then 1
    else (
        if (nvl($T_MP$,"1") = "1") then 1
        else 0))

[romaintailhurat]

Le premier JSON (srcv-buggy.json) correspond au questionnaire qui porte le bug, sa construction dans Pogues est la suivante:

En vert, le module qui contient les variables calculées (dont T_MENAGE_PRINC). En rouge, la séquence dans laquelle on affiche les variables (et dans laquelle on constate le bug).

Les deux modules qui suivnet TCM_THL_DET_PROXY ainsi que la séquence test sont dans des boucles liées à la boucle des prénoms (issue de TCM_THL_DET_PROXY).

Le second questionnaire (dans lequel on ne constate pas le problème) a la structure suivante - elle reprend l'idée du questionnaire problématique en le simplifiant:

[romaintailhurat]

Pour information: dans les deux cas, on utilise le module TCM_THL_DET_PROXY qui inclut le Pairwise.