Description
Two gaps we hit in the Kommunaler Weiterbildungsindex forms (formtransform 0.10.0, xlsform2lstsv).
1. parameters: randomize=true is not carried to LimeSurvey
For select_one / select_multiple with randomize=true in the parameters column, the DDI keeps it as a cdl:parameters note, but the LimeSurvey TSV has no answer-order setting. As a result, the order is randomized in KoboToolbox but fixed in LimeSurvey, so the same form behaves differently depending on the platform. We need randomization to avoid primacy effects in lists whose options are compared in hypothesis tests.
Expected: map randomize=true to the LimeSurvey question attribute random_order = 1 (list, multiple choice, and the filtered "most important" select_one that uses choice_filter). Ideally, options marked exclusive (e.g. "keine", "weiß nicht") and sonst/other stay at the end. If LimeSurvey can't pin them, a warning in the conversion output would help.
Reverse direction: random_order = 1 → parameters: randomize=true.
2. coalesce() is not in supportedXPathFunctions
coalesce() is a standard ODK XPath function and a common way to treat empty numeric fields as 0 in sum checks, e.g.
. >= 0 and . <= 100 and (coalesce(${a},0) + coalesce(${b},0) + .) = 100
At the moment the transpiler throws. Workaround: if(string-length(${a}) = 0, 0, ${a}).
Expected: transpile coalesce(x, y) to LimeSurvey EM, e.g. if(is_empty(x), y, x), and back.
🤖 Generated with Claude Code
Activity
fix(lstsv): optional questions with a constraint can be left empty; c…
feat(lstsv): carry randomize=true to LimeSurvey (#213) (#215)
jstet commentedon Nov 4, 2026
Both parts shipped in v0.11.0:
randomize=true→answer_order/subquestion_orderwithkeep_codes_order(#215), andcoalesce()(#216, together with #214). Pinning further options came later with thefixedchoices column (#237, #238, v0.12.0).