Found while investigating #2296
The problem is that we should not remove a directory locally if it contains
modified files.
But the modification time of the directory is not necessarily chaning (so
the instruction of the directory may still be NONE)
We have to move the child_modified test a bit down to be recursive
Squashed commit of the following:
commit 4d9b072f560fa171a1390b7c74425614aa20e955
Author: Olivier Goffart <ogoffart@woboq.com>
Date: Tue Oct 14 16:04:02 2014 +0200
Remove useless variable
commit 8e85de0307ec5f31bf3f92a7de793fed7d41c2ea
Author: Daniel Molkentin <danimo@owncloud.com>
Date: Tue Oct 14 16:01:52 2014 +0200
Make Windows Explorer Extension build
commit 8e2942cd9fd32e3af72d60cba0d06bd9d6222a45
Author: Daniel Molkentin <danimo@owncloud.com>
Date: Tue Oct 14 11:39:37 2014 +0200
Fix compilation
commit 0fc0c0e0e0c7e58ad97f62700256c7d1f8c0670b
Author: Olivier Goffart <ogoffart@woboq.com>
Date: Tue Oct 14 11:48:32 2014 +0200
Windows Shell Integration: Try to let the thread notify about changes when there are changes
commit 4a1712b7c03269ca3007f167b8f313ea47655967
Author: Olivier Goffart <ogoffart@woboq.com>
Date: Tue Oct 14 11:35:20 2014 +0200
Windows Shell Integration: Share the RemotePathChecker amongst all the OCOverlay instances
commit 2d87408e9af5a4d7ab71c460ce606ba1f367c09f
Author: Olivier Goffart <ogoffart@woboq.com>
Date: Mon Oct 13 18:55:15 2014 +0200
Windows Shell Integration: Attempts to wait on multiple objects (WIP)
commit e448e427b6d1561ad7a40d08fc6632f4d2b4ef44
Author: Daniel Molkentin <danimo@owncloud.com>
Date: Mon Oct 13 17:58:02 2014 +0200
Introduce a worker thread
commit 2344407ec0bc1ce173ebbacadcf3992d62d94078
Author: Olivier Goffart <ogoffart@woboq.com>
Date: Mon Oct 13 17:03:47 2014 +0200
Windows Shell Integration: try to keep the socket open using a thread (WIP)
commit ea6d5273ed60d8bc3f1c5d5c6936364d783a1c0f
Author: Daniel Molkentin <danimo@owncloud.com>
Date: Mon Oct 13 15:27:46 2014 +0200
Make Explorer plugin work again with named pipes
This is a temporary hack, which needs more refactoring.
commit 44a3437a44082379efa0078c9afd7b8bbde930de
Author: Daniel Molkentin <danimo@owncloud.com>
Date: Sat Oct 11 07:31:24 2014 +0200
Fix code
commit 123390a0f3516c0078309d7048c6d2acb9293676
Author: Olivier Goffart <ogoffart@woboq.com>
Date: Fri Oct 10 16:29:35 2014 +0200
Windows shell integration: Use named pipe (WIP)
commit 9eea7e2321abeac6b8db0bd85bfce612dbf6bb20
Author: Olivier Goffart <ogoffart@woboq.com>
Date: Wed Oct 1 12:04:13 2014 +0200
Windows Shell Integration: Simplify StringUtil
This fixes a memory leak in CommunicationSocket::ReadLine
It is possible that we have should_update_etag set to true for files
that we also need to propagate. In which case we must not write to the DB
too early as this could cause data loss. (cf: issue #2296)
In the sync engine. Because that makes tha tthe lower_bounds in selective sync works properly.
For example, if both "Test" and "Test Test" are in the list, then "Test/Foo" would match the "Test Test"
because slash is after space
Task #2289
The server keeps the mtime and etag anyway.
And changing the mtime also changes the etag and so make other clients re-download
the file instead of detecting the move
Issue #1385
When renaming, we would normaly already have an error if the file already exist.
And we want to allow the renaming to chang ethe case (Eg. "hello" -> "HELLO"
The problem was that the backup function failed as we tried to
move /home/xxx/ownCloud/ to /home/xxx/ownCloud/.oC_bak
The intention was to move it to /home/xxx/ownCloud.oC_bak
Issue #1989
When recovering from a inalid move operation, we call
avoidReadFromDbOnNextSync. This will remove the fileid from the db so
we don't detect a move on the next sync. But we want the next sync to fetch
the file id again, and this can only happen if we don't read them
from the db and do the actual PROPFIND, this is done by avoidReadFromDbOnNextSync
However, if there is propagation in that source directory later, it
will overide the invalid etag with the real one and we will not uissue the PROPFIND
to fetch the fileid.
We must therefore also protect the future write to the DB when calling
avoidReadFromDbOnNextSync
It is called automatically by QWizard and do not need to be called explicitly
But setVisibla(false) don't really have an effect there since show() is going
to be called on the page
When the url do not have a path (for example: "http://example.com" as
opposed to "http://example.com/"), its path is not a prefix of the
root path of the cookie (usually '/')
By adding the dav path, we make sure the URL has a path.
This made a bug when the owncloud url was just a domain name and did not
have a path
Otherwise the buffer might fill up too quickly and get too large and consume
too much memory which could lead to crash in extreme cases
Should fix issue #1974
Olivier Goffart