I am preloading a few identifying variables+2 collected data vars.
When I upload the TAB file with only identifying vars the assignments come through. When I zip the tab file and upload the zip-file I get the PL0000: Preloading data were not found. I checked my file again, data is there.
I simplified my questionnaire to the cover page with 6 vars, asked a colleague to try, it is always PL0000.
This is clearly reproducible.
Any idea what I can try?
possibly a wrong file name inside the zip archive.
check the name matches the same in the downloaded zip archive.
Checked again using the reduced version questionnaire of 6 vars:
Questionnaire name: trial
Tab file: trial.TAB
zip file: trial.ZIP
I downloaded the format: template_trial_v9.zip (but the version is not in the name).
Following this request, I couldn’t reproduce the above reported problem.
I will contact @lhuybreg directly to acquire the missing details.
ok let me know which color of list you want me to submit.
We gave up on preloading study variables, and uploaded identifying only. No solution found.
We found the issue: the file format “.TAB”.
When we save as “.tab” the error does not appear.
Dear @FMBARBA , thank you for reporting the issue.
The import procedure seeks files with particular names. I agree that the case-sensitivity in file names exists only in some, but not all operating systems.
For the moment, please make sure the filename in the zip-archive prepared for the preloading of assignments is exactly the same as the name of this file during the export of data produced with Survey Solutions.
If any files have been already prepared, please rename them, e.g. by running the following in the Windows command line (files must be in the current working directory):
ren *.TAB *.tab
or an equivalent command in your OS.
Thanks, Sergiy. Just to add that seems single uploads (“Identifying data only”) accepts both “.tab” and “.TAB”, while zip uploads (“Identifying and collected data”) only accepts “.tab” files (this is why it took us some time to figure it out).
Indeed. This is because for the single upload there is no need to infer anything from the name of the file, only from the content of the file, so it should be equally happy accepting *.txt files as well. But when uploading an archive, all the files in it are inspected, and the program tries to match them to the corresponding rosters in the questionnaire (by name). Hence the names become important, yet the case should be irrelevant.
I have noted this defect in the issue tracker and developers would know it under the ref.# KP-14709.
Since a workaround exists, this will be taken care of under the normal cycle of the development in one of the future releases.
I regret of the time lost due to this unclarity.