Hi there ,
operating system : Ubuntu 18.10; client 2.5.2rc2
i think title says a lot , simply follow the steps to reproduce the data loss :
regards
Also reproducable with a windows client. Ouch!
reproducible with Macos client as well , i have just tested it
Yes, simpler reproduction:
The sync engine sees this as a deletion of test.txt and the creation of an erroneous placeholder file other.txt.owncloud.
What's the expected behavior? Renaming test.txt to other.txt on the remote and dehydrating other.txt.owncloud?
I expect a similar problem to occur if one renamed test.txt.owncloud to other.txt: test.txt.owncloud would be recreated and a silly other.txt would be uploaded.
(Issue moved per @ckamm internal comment)
Reasons: This is complicated to fix. A fix would take a bit and probably not be in time for 2.5.3. Also, a fix in 2.5 would need to be redone against the new discovery code in 2.6, doubling the work. I think moving to 2.6 is acceptable since the feature is experimental.
@ckamm unfortunately it's still reproducible with ownCloud-2.6.0.11958.11762-daily20190510.msi

@lazawan On Windows with cfapi based vfs .owncloud files should have no special behavior at all. What exactly are you seeing?
@ckamm i added extension .ownCloud to a virtual file (a pdf file in virtual mode with Cloud sign ) and then this file was ignored from client and disappeared from server
Interesting, looks like .owncloud is in the ignore list? I'll try to reproduce.
@ckamm i have just checked the ignore list , .ownCloud ist not in the list
and i could reproduce it again :'(
@lazawan I've not been able to reproduce the problem with ownCloud-2.6.0.12048.11850-daily20190530.msi. When I rename to .ownCloud in my wincfapi vfs folder the rename gets propagated to the server.
The behavior you describe would happen for ignored files. Could you double check your C:\Users\$USER\AppData\Roaming\ownCloud\sync-exclude.lst? Some in-between test versions when we started working on vfs added the .ownCloud extension there.
If it's not there: could you get me some logs?
macOS 10.14.5, 2.6.0alpha2 (build 12128)
Scenario 1
Scenario 2
I can reproduce the issue in scenario 2 - which is odd because there's an explicit autotest for it and that passes.
Works as expected on macOS 10.14.6, client 2.6.0beta1 (build 12241)
Most helpful comment
Reasons: This is complicated to fix. A fix would take a bit and probably not be in time for 2.5.3. Also, a fix in 2.5 would need to be redone against the new discovery code in 2.6, doubling the work. I think moving to 2.6 is acceptable since the feature is experimental.