| Age | Commit message (Collapse) | Author |
|---|
|
Closes #123 (room account data were parsed even before). No specific event classes for account data yet, though.
|
|
This fixes compilation with older compilers that try to instantiate a copy constructor (and fail because unique_ptr) but actually is more proper in general. Also: do not advertise the class as Q_MOVABLE_TYPE: this was useful for QList/QVector when SyncRoomData was copyable; now it isn't, and Qt containers can't deal with non-copyable classes at all.
|
|
This causes the following changes along the way:
- Owning<> template is decommissioned.
- event.h has been rearranged, and Event/RoomEvent::fromJson static methods have been replaced with an external makeEvent<> function template. A side effect of that is that one cannot use a factory with a type other than the one it's defined for (i.e. you cannot call makeEvent<TypingEvent>) but that feature has been out of use for long anyway.
- Room::doAddNewMessageEvents() and Room::doAddHistoricalMessageEvents() have been removed, giving place to Room::onAddNewTimelineEvents() and Room::onAddHistoricalTimelineEvents(). The most important difference is that all code that must be executed now resides in addNewMessageEvents() (it moved from Room to Room::Private) and classes inheriting from Room are not obliged to call the overridden function from the overriding function (they can do it but those functions have empty bodies in Room). This was a long overdue change, and owning pointers simply mandated it. Room::onAddNewTimelineEvents/onAddHistoricalTimelineEvents should not do anything with the passed range in terms of ownership, it's just a way to allow the derived class to update his data in due course.
- Room::Private::dropDuplicateEvents() and Room::Private::insertEvents(), notably, have been updated to work with owning pointers. insertEvents() move()s pointers to the timeline, while dropDuplicateEvents uses remove_if instead of stable_partition and doesn't explicitly delete event objects.
Also, a bugfix: Event accidentally had not virtual destructor for quite a long time. According to the standard, deleting an object through a pointer to a base class without a virtual destructor leads to UB. So the fact that libqmatrixclient clients even worked all these months is mere coincidence and compiler authors good will :-D
|
|
|
|
Having to pass ConnectionData to each and every job class was nothing but boilerplate since the very beginning. Removing it required to prepend BaseJob::start() with ConnectionData-setting code, and to provide a way to alter the request configuration depending on the (late-coming) ConnectionData object. This is a new responsibility of BaseJob::start(); the previous BaseJob::start() contents have moved to BaseJob::sendRequest() (which is now invoked on retries, instead of start()).
|
|
|
|
Kicking and inviting use generated job classes. Rooms in Invite state are stored separately in the hash from those in Join/Leave state because The Spec says so. For clients, this means that the same room may appear twice in the rooms map if it's been left and then the user was again invited to it. The code in Quaternion that properly processes this will arrive shortly.
|
|
|
|
|
|
|
|
|
|
The biggest change is we have no pimpls in Event objects anymore - because it's two new's instead of one per Event, and we have thousands and even more of Events created during initial sync. The other big change is introduction of RoomEvent, so that now the structure of events almost precisely reflects the CS API spec. The refactoring made UnknownEvent unnecessary as a separate class; a respective base class (either RoomEvent or Event) is used for this purpose now. All the other changes are consequences of these (mostly of RoomEvent introduction).
|
|
After adding some profiling it became clear that to recalculate the room name and emit namesChanged() upon each member event is a waste, especially when there are thousands of those coming at initial sync (*cough* Matrix HQ room). So the room name is recalculated only once and unconditionally (in most cases this will boil down to checking whether name/canonicalAlias changed after processing the events batch), and namesChanged is only emitted once per batch, if any name or alias changed.
|
|
Mainly it's about const-ification (in particular, passing const-refs instead of values) and deleting unneeded declarations/#includes. Since the changes alter the external interface, this is submitted as a PR for peer review.
One of unneeded declarations/definitions is a virtual destructor in BaseJob descendants. Since a job object should be deleted through QObject::deleteLater() anyway (and it's the only correct way of disposing of the object), all deletions will call the stack of destructors through virtual QObject::~QObject(). Therefore even BaseJob could get on with a non-virtual destructor but for the sake of clarity BaseJob::~BaseJob() is still declared virtual.
|
|
there
Because these fall outside of SyncJob and Event context, respectively. In addition, Owning<> has gained a move assignment operator (because we have a move constructor) and assign() convenience method to take ownership over an existing container; also, Owning<>::release() is done the right way now (the previous version was copying the return value to a new container instead of releasing the old container).
|
|
|
|
|
|
data() in each job class
|
|
|
|
|
|
|
|
Replaced QList<> with QVector<> where appropriate + minor code cleanup
|
|
Upon discussion with @Fxrh in #quaternion, this now comes in master,
|
|
See https://marcmutz.wordpress.com/effective-qt/containers/ for the background and http://lists.qt-project.org/pipermail/development/2015-July/022283.html for the relevant flamewar in Qt dev mailing list.
|
|
Removed unused #includes, fixed incorrect #endif comment.
|
|
return it
This better fixes the contract for derived job classes and simplifies error reporting. Methods error() and errorString() are kept for back-compatibility; status() returns a combination of them, conveniently packed into a Status object. For a quick status check, Status::good() is provided.
|
|
|
|
data()
Thanks to CLang model.
|
|
actually introduce EventList class)
This makes loading of room events more compliant with the spec, not trying to load from keys that are not supposed to be there. As a result of refactoring along the way, a dedicated QList<Event*> subclass is made that remembers the JSON key it should load from and actually load itself from a QJsonArray.
|
|
|
