summaryrefslogtreecommitdiffstats
path: root/synchronizer
Commit message (Collapse)AuthorAge
* SIGSTOP is not avialable on windowsChristian Mollekopf2018-04-19
|
* Windows compatChristian Mollekopf2018-04-19
|
* Windows compatChristian Mollekopf2018-04-19
|
* Windows compatChristian Mollekopf2018-04-19
|
* Setting the dictionary value works.Christian Mollekopf2018-04-04
|
* Try again to disable the dockChristian Mollekopf2018-04-04
|
* Try againChristian Mollekopf2018-04-04
|
* Hide the QGuiApplication on mac os again.Christian Mollekopf2018-04-04
|
* Revert "The QGuiApplication does not seem to be necessary for QTextDocument."Christian Mollekopf2018-04-02
| | | | | | | This reverts commit 630592544d719fdda8e4c942329492d1a41a8b14. Depending on the mail we do start crashing. QTextDocument sometimes ends up using the palette which relies on an initialized platform plugin.
* The QGuiApplication does not seem to be necessary for QTextDocument.Christian Mollekopf2018-03-18
| | | | ..as long as we link to QtGui.
* Xapian based fulltext indexingChristian Mollekopf2018-02-11
| | | | | This cuts into the sync performance by about 40%, but gives us fast fulltext searching for all local content.
* Dump some process stats on exit in the synchronizer.Christian Mollekopf2018-02-05
|
* Support for storage upgradesChristian Mollekopf2018-01-30
|
* Use imported targets instead of qt5_use_modulesHeiko Becker2017-07-29
| | | | | | | | | | | | | | | | | | | From Qt's documentation: "This macro is obsolete. Use target_link_libraries with IMPORTED targets instead." It's only recommended with cmake >=2.8.9 & < 2.8.12. Sink already requires cmake 3.0. One advantage of using the imported targets is, that cmake complains if a target isn't found before it's used, like Qt5Concurrent missing from the find_package_call here. Reviewers: #sink, cmollekopf Reviewed By: #sink, cmollekopf Subscribers: #sink Tags: #sink Differential Revision: https://phabricator.kde.org/D6361
* Link to qt::core as wellChristian Mollekopf2017-06-06
|
* No more SINK_DEBUG_AREAChristian Mollekopf2017-05-12
|
* Don't store the lockfile inside the main resource folder.Christian Mollekopf2017-03-26
| | | | | | | | | We currently use the folder as the main db folder, meaning we remove the folder when removing the db. This results in the lockfile vanishing with the db, which then confuses QLockFile (resulting in a lot of warnings). We may want to start moving everything into a resource-instance folder, but then we have to do it properly across the board.
* Adapt to KAsync changesChristian Mollekopf2017-03-01
|
* Make sure we print the message.Christian Mollekopf2017-02-23
| | | | std::cout doesn't seem to work reliably at this point...
* Print more info why acquiring the lock fails.Christian Mollekopf2017-02-23
|
* Feed qt debug messages into the same stream as the rest.Christian Mollekopf2017-01-23
|
* Directly linking against dl doesn't work on freebsd.Christian Mollekopf2017-01-23
| | | | Patch by Tobias Berner.
* gdb mode that can be used to attach a debuggerChristian Mollekopf2017-01-18
|
* Make sure the path is existing, otherwise the lockfile will not beChristian Mollekopf2017-01-10
| | | | created.
* Create the lockfiles in the data directory.Christian Mollekopf2017-01-09
| | | | The ended up in the home directory.
* We don't require qt widgets.Christian Mollekopf2016-12-13
|
* Debug outputChristian Mollekopf2016-12-06
|
* Sleep on crash so we have time to attach a debugger.Christian Mollekopf2016-09-15
|
* CleanupChristian Mollekopf2016-09-15
|
* A new debug system.Christian Mollekopf2016-07-07
| | | | | | | | | | | | | | | Instead of a single #define as debug area the new system allows for an identifier for each debug message with the structure component.area. The component is a dot separated identifier of the runtime component, such as the process or the plugin. The area is the code component, and can be as such defined at compiletime. The idea of this system is that it becomes possible to i.e. look at the output of all messages in the query subsystem of a specific resource (something that happens in the client process, but in the resource-specific subcomponent). The new macros are supposed to be less likely to clash with other names, hence the new names.
* Prepared sinksh traceChristian Mollekopf2016-07-07
|
* Catch errorsChristian Mollekopf2016-06-21
|
* Disable the quitlockChristian Mollekopf2016-06-05
| | | | | Without this the synchronizer process regularly closes prematurely, not sure why.
* Print a demangled stacktraceChristian Mollekopf2016-06-03
|
* Log the resource start.Christian Mollekopf2016-05-30
|
* Propagate test-mode to resources.Christian Mollekopf2016-05-23
| | | | This way we can run tests in completely separate data-directories.
* A better crashhandler that notifies the client of the crash.Christian Mollekopf2016-05-22
|
* Don't encode the resource type into the identifierChristian Mollekopf2016-04-17
|
* Fromatted the whole codebase with clang-format.Christian Mollekopf2016-03-03
| | | | clang-format -i */**{.cpp,.h}
* Use Sink instead of SinkCommonChristian Mollekopf2016-02-09
|
* Introduced debug areas and prettified output.Christian Mollekopf2016-02-04
|
* Renamed Akonadi2 to SinkChristian Mollekopf2016-01-20
| | | | (except for documentation).
* Stacktrace printing on crashChristian Mollekopf2015-09-10
|
* Set stale lock time.Christian Mollekopf2015-08-10
| | | | | | We're trying to protect against race conditions, a lock time off 500ms should suffice and means we can recover automatically after a crash.
* Moved lockfile into synchronizerChristian Mollekopf2015-07-30
| | | | For better testability of listener.
* Moved listener to commonChristian Mollekopf2015-07-28
| | | | So we can use it in tests as well.
* Don't set the current revision upon start.Christian Mollekopf2015-07-27
| | | | The client can simply read it from the database.
* Immediately close the socket after a shutdown notification.Christian Mollekopf2015-07-23
| | | | | | | | Otherwise it can happen that the synchronizer receives the shutdown command from a ResourceAccess instance, and while the synchronizer waits for quit to be called, a new ResourceAccess instance already connects to the synchronizer. That new connection will then of course be unexpectedly killed once quit is called in the synchronizer.
* Fixed buildChristian Mollekopf2015-07-23
|
* Avoid slots being called after the clients have been cleaned up already.Christian Mollekopf2015-07-23
| | | | | It is possible that the clientDropped slot is called when we close the socket, so better disconnect everythign first.
generated by cgit v1.2.3 (git 2.25.1) at 2025-04-28 20:32:27 +0000