* DBus mixins moved to dbus.h
* Json mixins moved to json.h
* Icon mixin moved to icon.h
* Comparison mixins moved to compare.h
* Hash and string mixins moved to blackmiscfreefunctions.h
* Index mixin moved to propertyindexvariantmap.h
* MetaType mixins moved to variant.h
* registerMetaValueType moved to variant.h
* valueobject_private.h renamed to variant_private.h
* Involves rearranging some header includes to break cyclic include dependencies,
* Adding a new, simple base class CEmpty,
* Removing any remaining polymorphic uses of CValueObject with templates,
* Adding a new trait for use with enable_if to restrict templates to work only with value objects,
* Replacing the polymorphic/runtime multimethod-based compare functions with static/compile-time compare functions.
* GUI component for aircraft parts
* remote aircraft selector component
* Adjusted GUI for internals component
* Enable / disable debug messages from GUI
* Allow to init engines directly
* Removed unused async sort in sequence
In same step fixed found issues in interpolator
* allow to set max rendered aircraft
* extended unit tests for interpolator / parts testing
* allow to skip sorting when splitting by time
* update ot aircraft to FSX in own member function
* Skip time sync parts (FSX) when disabled
(based on FSX testing)
* Only send changed situation to SIM
* Split sending of parts / situations
* Only send parts with a reduced frequency (means slower as positions)
* Mark geodetic height as null for default values (the value is usually unavailable)
* Fixed altitude to MSL for network data
* Trace which aircrafts support aircraft parts via network
* Renamed insert_fron push_front (as proposed by Roland)
Status quo / lessons learnt
* On slower PCs jitter is still noticed for interpolated aircraft.
* Running interpolation in an independent process (aka core, not in GUI) reduced load dependencies
=> it seems to make sense to run driver in own thread
* The onGround flag in parts seems clumsy as it required to retrieve parts for position updates
* In interpolation performance really matters
* using split by callsign everywhere
* helper function to insert_front
* revised linear interpolator
* renamed to remoteAircraft
* renamed to container() in providers (gettters are usually copies)
Issues why changes did so long:
* insert in list is not adding in front, but same as push_back (that was confusing)
* naming of values before/after in interpolator was ambigious
* QMap keeps values sorted by key, not arbitrarily
* The map will be used with CSimulaorInfo to transfer a variable number of setting information about the flight simulator. It is being used, because it is already DBus and tupel compliant. The new name better fits its purpose.
* Add == operator for CIndexVariantMap (ambiguity error with tupels)
CSequence uses std::find, and CCollection uses the more efficient find method of the implementation type,
which requires that the implementation type be an associative container like QSet or std::set.
Modified CContainerBase::contains to use these new find methods.
for consistency with the container classes and to avoid any potential infinite recursion with MSVC2010.
See also commit:6a9065b
Also fixed a mistake in CSequence and CCollection, where Pimpl::erase was calling the wrong version of m_impl.erase
reported by Roland.
