The problem occurs because of the sorting of items when we have things like
bigfolder
bigfolder/bigsubfolder
bigfolder-2
Then, because dashes come before slash in ascii, the bigfolder-2 would come before its content
and the propagator would thinkg bigfolder is empty and save the etag before it is processed
Should fix issue #2832
That makes it defaulting to the APPLICATION_NAME which is ownCloud.
That is what we want and expect to read the system exclude file from
on linux.
How we find the system exclude file linux:
ConfigFile::excludeFileFromSystem()
-> Theme::instance()->appName()
-> OwncloudTheme::appName() returns "ownCloud"
-> default Theme::appName() returns APPLICATION_SHORTNAME
-> cmake installs to ${SYSCONFDIR}/${APPLICATION_SHORTNAME}
It seems like verclsid.exe hangs on our class IDs when invoked
through explorer.exe for 5-10 seconds. It doesn't hang if I
invoke the same command line from cmd.exe, so there could be some
process parameters that don't play well with our extra thread
or to the pipe connection that we do in it.
Delay creating the RemotePathChecker thread until the first
IsMemberOf call. verclsid.exe only seems to instantiate a object
of each registered class, without actually using them, so we
can use this as a workaround.
This should be fixing issue #2680.
* Fix various spaces vs tabs issues
* Rewrite
- to use instance variables consistently
- to not make shared state explicit instead of relying on
class variables of a common base
* Fix reconnecting to an oC client.
Occasionally on Windows ~DiscoveryMainThread() triggered a crash.
Probably because the QMutableLinkedListIterator was destroyed after
it's underlying data had been removed (from the map).
Cleaning up the map and lists was redundant anyway, so the
destructor is changed to only explicitly destroy the vio_file_stat
objects.
The _listeners list is destroyed before the _localServer.
And since _localServer is the parent of all generated QLocalSockets,
they get destroyed in turn - which triggers onLostConnection() and
thus accessed the destroyed _listeners list.
To avoid that, delete all active QLocalSockets in SocketApi before
its members are destructed.
We also now delete sockets when we're done with them. I think
disconnected sockets would otherwise linger until SocketApi destruction.
