IQSS
Another Duplicate File Issue: Stuck in the middle with you.
See RT 295094, Julian mentioned some intermittent issue uploading files where it failed with the dupe files message:
The first time I tried to upload both files at the same time by dragging them both into the window, after the progress bar for both files filled up, I got an error message about a duplicate file name "0V". I thought it would happen again and planned to take a screenshot then, but it didn't happen when I tried uploading the files a second time.
When I hit cancel and tried the second time, I dragged one file in first, waited for it to load, which took a few seconds, and dragged the second file in. And I didn't get an error for either file. So I tried dragging both files in so I could copy the duplicate file name error, but both files uploaded successfully.
I hit cancel again and asked the depositor to try uploading both files again.
There are 68 files with the file name "0V".
The dataset, doi:10.7910/DVN/OJHYZA has 68 duplicate files named 0V. This likely occurred through a loop hole in an earlier version of Dataverse, maybe 5.0.
There are also 2 duplicate files named 0.5V. There may be others. Not sure whether fixing them entails only renaming them in a new draft or in the existing published version since it is an illegal state and may cause other issues at a later date.
The reason this became an issue is that you cannot upload or save any changes to files until these dupes are fixed.
- Upload 2 new files, save, fails with error: × The files could not be updated. – Duplicate filenames: 0V
- Rename all 68 0V files in UI, save, fails with error: × The files could not be updated. – Duplicate filenames: 0.5V
This likely occurred through a loop hole in an earlier version of Dataverse, maybe 5.0.
No, definitely before v5 - I remember that dataset, it had other issues. :)
The fact that you can't even upload new files if there are duplicate filenames is pretty insane. And as for fixing any existing dupe filenames - it just sounds like they need to be fixed all at once...
(I'll handle fixing it in the specific dataset above; let's use this issue for any code fixes)
Hmm, I haven't been able to reproduce it just yet. I created a dataset, and created a few duplicates (in the db directly). And so far I can both upload, and change 2 dupe filenames at a time... Do I need a published version maybe? (the dataset above does)
I don't know what's going on. I was able to reproduce the above condition when I published my test dataset, and then (immedately) tried to upload another file. Then it failed to save with a dupe message, complaining about an existing duplicate. But once I have created a draft, by making a metadata change, I am back to being able to make any changes I want.
Uh, including being able to create new duplicates. Something is still seriously buggy there. I'll let somebody else try to figure out what's going on.
Last note: With the dataset above, it WAS possible to fix all the dupes in the UI; without having to mess with direct database updates. They just had to be changed all at once. Trying to change just some was resulting in an error. IDK - is this a bug, or is this how it was supposed to be implemented? - I'll leave it to those who worked on on this functionality recently.
I'm still curious as to why I can't reproduce this on my own dev. box; what's so different between my dataset and the one above? edit: It's not just that I can't reproduce the above; for me the duplicate name check is just no longer being enforced at all, once I have an existing draft version. Hmm?
@landreev this might help you reproduce the issue:
- https://github.com/IQSS/dataverse-sample-data/issues/32
Or maybe what I found is a different issue? I'm not sure.
