Replacing/updating an existing file

[atrisovic]

Hi there,

DVUploader won't detect that a file has changed unless I use the -verify flag.
However, when I use it, it sees and uploads the file as a new one instead of replacing (ie updating) an existing file. So the dataset will have a new text-1.txt file instead of updating the test.txt file.

Thanks!
Ana

[behinger]

bump I have exactly the same issue. I wonder if there is an interaction with the versioning system - i.e. doesn't work in draft mode but works in published mode?

[qqmyers]

Unless I'm forgetting something, DVUploader doesn't currently understand that Dataverse has a replace function. It detects whether a file exists in the dataset already by path/name or, with -verify, by hash value and, if the local file is different, it uploads it, which then invokes Dataverse's renaming logic to avoid the conflict. It probably could use the replace API instead. Not sure if it should default to that though as replace in a draft version completely removes the old file which may not be intended - that could be addressed with another command line option though.

I'm not sure when this will be a priority (pull requests welcome) , but I can think of a few partial workarounds - you can detect the changes and use -listonly to not upload them and then manually address those files (either calling replace yourself or using the UI, or if you don't care about whether it is recorded as a file replacement, just deleting the file in Dataverse before reuploading with DVUploader).