I can reproduce the issue - screenshot and sample token export file are attached.

This is not really a bug but rather a feature request as it was not implemented at all.

Fixing the column names at export time would break token-import as only attribute_1...attribute_n fieldnames can be imported.
So this would also need a fix for token import:

• we should make sure that the 'Real Name' choosen by the survey admin is compliant with CSV field (we may need to escape it for CSV at export time),
• we would need to update the survey table at token import because this is where the token description are recorded
• we would need to cope with multiple import of CSV files that may not have the same token attributes columns: in this case we should decide what to do for token description (override, merge...).

This is not impossible, but is clearly a feature request and not a bug.

Matador,
I can't see the Feature Request project anymore ?
Where to move this ticket then ?

TIA,
Thibault

Thibault, thanks for your feedback. From the user point of view it's really a problem if your columns are named "attribute_1, attribute_2, ..." after export because when having several columns you might not realy know what data is stored at each column.

I suggest to:
a) export

• Export columns names

b) import

• Check if the column names of the CSV file exist in the token table. If so import them according to the headings. Attribute description in survey table should match import file headings.
• If the heading names do not match show a warning and import them in the order they are listed at the import file. we should not override the attribute description.

Mzi, your proposal would add extra complexity to token import.

I propose a simpler solution:

a) export

• Exports attribute_x code with an extra <attr description> text

b) import

• strip any <attr description> and import a usual.

See rev 7956.

Setting this as resolved, please re-open if it's really not satisfactory.