Opened 17 months ago
Uploading empty directory results in nonsensical items listed in failed transfers
|Reported by:||php4fan||Owned by:|
|Component version:||3.51||Operating system type:||Linux|
|Operating system version:||openSUSE Tumbleweed|
- Upload a folder that contains some empty folders
- Upload it again
- the first time, if the subfolders don't already exist on the server, they will be created
- the second times, no folders will be created on server as they already exist.
Nothing should be listed in "failed transfers" as no transfer has failed
After the second time (or already the first if any of the empty folders already existed on the server already), the empty folders will be listed in the "Failed transfers" tab.
That makes no sense whatsoever because you don't upload an empty folder, you create it on the server. When "uploading" a folder, if it already exists on the server, you either know that by requesting the listing beforehand and hence shouldn't have to attempt creating it, or you try to create it anyway and detect that it already exists because you fail to create it. Either way, that's not an error and it shouldn't be listed as a failed transfer. The only case where a folder should be listed as a failed transfer is if you fail to create it on the server but it does not already exist.
- the "reason" column says "Creating directory". That makes no sense either. You are saying that you failed to upload a directory because "creating directory".
- When I have failed transfers, usually what I do is: reset and requeue all, and then process the queue again. This results in the folders showing up again in failed transfers. So, it looks like there was some error creating the folders on the server and that errors keeps happening and cannot be fixed. And none of that is true: there is no error to begin with, the folders simply already exist.
Note that if the folders contain files, none of this happens. All is uploaded as expected and nothing is listed under failed transfers. As expected.
I'm setting priority as high because false errors are as bad as failing to detect real errors. If an error log of any kind is polluted with false positives that I have to ignore, I'll end up missing real errors.
I think this is a regression, I hadn't observed this until recently.