For now it is available in the positioned Nasal
module. FGAirport is fully ported, but especially
procedures/navaids and free functions are still
missing.
Iterator must be reset after clearing waypoint list.
Deleting "*waypoint.end()" doesn't delete the last waypoint (end()-1
does - it is never actually used though).
PositionedID is of "int64_t", which depends on platform: apparently it is
"long long int" for Mac (requiring %lld format), but it is "long int" for
64bit Linux (requiring %ld). To avoid a "compiler warning fix commit
war" ;-) use a type cast to the longest common type (long long int).
Inspired by Harald's patch, attempt to get AI models show realistic gear animations. Unfortunately AI models seem to be ignoring the gear properties, so keeping the code disabled until I can trace down why.
Track length is calculate every iteration/for every aircraft. Since some
tracks have hundreds of waypoints, calculation is slow. Cache the
track length between the waypoints ahead, so it rarely needs to be
recalculated.
AIManager had unfortunate code to pass some state to the traffic-manager; this is no longer required since traffic schedules hold onto their AI-Aircraft directly as needed.
Make landings and takeoffs look more correct; tweak climb-out and touchdown phases in particular, so the turn to destination heading occurs earlier on climb out, and touchdown occurs close the GS transmitter / some distance down the runway from the beginning.
Expand the performance DB logic to support aliases, and select based on aircraft type as well as class. This allows to introduce some variation into AI traffic performance. Change the initial climb-out waypoints to use pitch-hold until passing 3000', which looks much more convincing
In preparation for caching the groundnet in the NavCache, make taxi-nodes and parkings inherit from FGPositioned. As part of this, make them heap (as opposed to value) classes, disable their copy-constructors, remove many mutating operations, and give them real constructors.
Also adapt MP and traffic module to enable the AI module when required.
This makes /sim/ai/enabled an internal property: it can default to false
(in future), and can be enabled as soon as any user-level feature (traffic,
MP, local weather, ...) requires it.
SceneryPager singleton must not be removed while FGScenery is still alive,
so hold a reference to it in FGScenery, saving the pager from being
deleted first.
distance_to_go / lead_distance were not initialized in all constructors.
Fixes 'condition on uninitialized data' reported by valgrind.
Also comment-out unused vars.
MP aircraft are loaded by a separate OSG thread (introduced after FG2.4.0).
The OSG thread also calls the "modelLoaded" callback. However, we mustn't
allow the OSG thread to call FGNasalModelData::modelLoaded directly:
FGNasalModelData isn't thread-safe. There are obvious issues (_callCount++;),
tricky issues like calling the Nasal _parser_ or creating hashes and
modifying the global Nasal namespace. It doesn't use locks to protect
against another thread executing a Nasal context or running garbage
collection. It also executes Nasal code itself (the model's "load" hook),
which we also cannot allow in a separate thread...
This patch returns all Nasal parts of MP-aircraft loading (parsing,
module creation, execution) to the main thread, while keeping the
multi-threaded OSG part (loading of MP-aircraft model files itself).
The same issue exists with scenery models (see other commit).
To summarize with 2 words: It s*cks... ;-)
* Pregenerating taxiroutes could interfere with runway assignments by ATC, when conditions changed, resulting in a taxi to one runway and a takeoff from another
* A simpler solution for the "Error in Traffic record bug". This still needs some more testing, but I haven't seen any error message anymore, since changing to the current code.
* Initialize AI traffic at speed zero, this should prevent some weirdness observed under boundary conditions.
* Don't activate groundnetwork proximity detection for pushback traffic until the "ready for startup message is transmitted. this should allow sufficient time for them to reserve a route, but a little more testing still needs to be done.
Replace SG_GENERAL by more specific log classes in many places.
Allow "," to separate logging classes (using "|" is odd on the
command-line).
Also add new option to make logging more useful for developers.
You can use:
--log-level=debug --log-class=environment
to only get environment debug messages, or
--log-level=debug --log-class=sound,ai
to only get debug messages related to the sound or AI subsystem.
Untangle AI reinit and init methods. Some code in init hooks expects to be
called once only. Derived classes should not redirect their reinit to init,
even if this seems fine for the methods of the derived class itself. This
also triggers the init methods of all base classes, which may not expect
multiple calls to their init methods (or to "init" on "reinit").
FGFX objects must be dereferenced early enough, and sound manager must be
removed late enough - otherwise openal complains about resources being
still in use when tryin to remove buffers.
Also: do not create FGFX objects for AI/MP aircraft, when AI sound is
disabled.
* Improved groundnetwork routing algorithm. Don't uncesscarily block taxiways. Instead, use a "just-in-time" blocking system. The unblocking algorithm still needs some work, but the current version is already a major improvement over the previous version.
* Some tweaks to the handover from ground to tower controller. Aircraft could refuse to take-off of even refuse to taxi onto the runway. This now seems to be solved.
* Landing AIAircraft now land properly!!!
* Fixed problems with takeoff scheduling.
* Untangled several vertical speed calculation algorithms.
* No more bent-over-backwards and jump-in-the-air AI aircraft initializations.
* Fixed intermixed target speed and target altitude during climb phase.
* Implemented beginnings of a "line up and wait" procedure.
* Tried to use actual runway elevation for calculating the decent profile.
Traffic Manager:
* Just continue routing until we run out of flights. This change removes one of the major requirements for setting the "Home port" field.
* Add a time restriction requirement for the aircraft scheduler; this became necessary after removing the limited-to-home-port routing restriction.
* Added a new field to the heuristics calculation: take into account whether an aircraft has already been used in a previous session. Rotate aircraft assignments for greater variability across sessions.
* Added a revision number to the cache files, so that old cache results, which are no longer compatible with the new file format, are discarded.
Groundnetwork and traffic control:
* Added a revision number to the cache files, so that old and incompatible results are discarded.
* The caching algorithm probably didn't store the correct data for airports that were processed while the user was quite far away. This is now corrected by checking whether the cached elevation data are equal to the generic airport elevation.
AIAircraft:
* I've been searching for the infamous aircraft bend-over-backward bug, that can occur during initialization, but to no avail yet. The only variable potentially responsible (tgt_vs) wich can explain the irregular jumping behavior, as well as the weird pitch results is initialized in AIAircraft's only constructor (through AIBase), and I can't find any situation in the ground handling code where this variable could get bizarre values. But,
* a couple of tgt_vs. calculations appear to be completely redundant. This value was calculated twice inside the ProcessFlightplan function, and subsequently again in the updateSecondaryTargetValues function. I have removed the calculations in the process flightplan function, without any apparent side effect.
Add a LOD (range animation) to each AI model if the property
/sim/rendering/static-lod/ai is set to any number greater than
zero. AI models get loaded but only displayed when closer than
the number given in the property (in meters).
If the property is absent or it's value is less or equal 0.0
than no LOD is added to the AI model.
- Better transistion between taxi and takeoff phases.
- Skipping the pushback stages when a gate doesn't require push-back.
- Some test code always chooses the user aircraft as the one that has to hold position.
- Change AIFlightPlan::waypoint into an independent class FGAIWaypoint
- Don't update leg count until the aircraft has passed the last waypoint that was created by the corresponding leg
- Clear ground network rendering for controllers that are no longer active.
--ai-scenario=... can only add/enable another scenario.
Introduce --disable-ai-scenarios option to disable all scenarios (can be
used by external launchers, GUIs etc)
Also provide error instead of debug message when a scenario cannot be loaded.
Move the disable-HOT feature from the MP aircraft to the a/c base class,
disable HOT for all AIaircraft, since that's a fix/work-around for
#242: AI aircraft respect the user a/c only when HOT is _disabled_ for them
#221: AI aircraft don't stack at parking positions when HOT is disabled
Also generally disables HOT for ballistic and other models (suggested by vivian),
allowing it for ship/carrier models only.
* Update AI aircraft object based on user position.
* Process flightplan
* Don't try to update the radar information for AIAircraft that don't have a valid manager pointer
- Fixed a bug in AI aircraft ground steering code: When aircraft were not moving, the value of headingchangeRate kept increasing to insane levels. Although this was clamped to a maximum of 30 degrees per second, the initial rate could still push the aircraft in the wrong direction. In practice, this bug would be visible when an AI aicraft would be pushed back, when it tended to veer to the right.
- Make sure that the aircraft slows down well ahead of the pushback point. This change ensures that the AC will actually reach the pushback point. It also ensures a slightly tighter steering range.
- AI ground steering rate is tuned to 30 degrees per second at a nominal taxispeed of 15. I now modulate the heading adjustment rate by manipulating the adjustment using a non-linear function (the sqrt). This allows for a slightly tighter turn radius at speeds < 15 and slightly looser turns at speeds > 15.
- The AI Flightplan generation code can return false. This can be used to determine whether any additional AI aircraft may be created. Currently, the function returns false when no more parkings are available. This should limit the build-up of huge AIAircraft tower stacks.
- The ground network can now graphically display all aircraft actitivy on the ground network by using a virtual marker system.
* Added an AIAircraft that is shadowing the user aircraft's behavior, in terms of AI / ATC interactions
* Set up a shadow AI FlightPlan that is used for tracking the user's aircraft.
Almost all FGPropertyData elements received via MP were leaked.
Property data is now cleanly deallocated in the FGExternalMotionData destructor.
Thanks to Jester for reporting rising mem consumption in MP mode.
Priority scheme for tile loading/removing
Cleaner tile manager interface for AI/groundcache/...
Reduce start-up delay. Drop splash screen when inner scenery is available.
Be nice to osg loader: slow-down main loop while initial scenery is still loading.
* New features
- More realistic descent paths
- Separation during descent and approach
- ATC approach controller (still silent)
- inbound traffic flow will start immediately
* Bug fixes
- Properly handle vertical speed when on ground
- Departing aircraft now wait for taxiclerance before moving
- Traffic manager waits for proper weather initialization
- Fixed instabilities in the preferential runway usage code
- Fine tuning of waypoint following code.
A small patch to allow the user to ignore specific
MP aircraft on a per-MP instance basis by setting
"controls/invisible". This
property name was chosen for consistency with the existing property
for AIBallistic objects.
issue is numerical problems in _wgs84_inverse, triggered by the traffic-manager
testing enormous distances geodetically.
Changes
- refactor AISchedule::update, so the aircraft approximate position
calculation is simpler.
- refactor AIAircraft to no longer use exceptions as a return value mechanism
- change AISchedule::update to use cartesian math for in-visible-range check
Improvements to the cloudsystem:
- A new xml format
- Texture indexing based on the position of the sprite in the cloud
mass, allowing more control over the texture set.
- Improved fog and shading
- Better sprite distribution
- A more natural distribution of clouds, so no more obvious grids.
This is now included in the model xml file.
So, wires and catapults also work even if they are not loaded by an AICarrier.
Modified Files:
AICarrier.cxx AICarrier.hxx
The current code still has some rough edges, in particular memory still
needs to be deallocated where possible, and the actual use of the code
needs more testing. This code has been running without noticable problems,
so I think it's ready for some wider exposure. Detailed changes include:
- Finetuning of the SID/STAR data concept.
- Preloading of all SIDs, from one xml file.
- ATC determines which SID should be used and echoes this over the com1 or
com2 radio.
experimental: Only one SID per runway is supported, and the waypoints are
read from file at every request, which is not very efficient. The current
code is only executed when FlightGear is configured to use airport dynamics
data from the scenery repository, instead of the base package. Since the
latter is still the default, I believe that this will provide enough safe-
guarding to commit SID/STAR support in small, incremental steps.
Next step will be buffering and support for multiple departure routes per
runway.
* Changed the runway XX. ATC message to actually report the real
designated departure runway
* In case of multiple active runways, select the one with a heading that is
closest to the direction of the ultimate departure destination / lines up
with the arrival path.
* Some support for geometry information provided by the custom scenery
project. Current support is for AI groundnets and runway use files only
since this is a switch that involves a lot of data verification and
updating, during the transistion the actual path where the data can be
read from is user configurable. setting the property
/sim/traffic-manager/use-custom-scenery-data to true
will cause flightgear to read the ground networks from the scenery
directory (--{fg-scenery}/Airports/[I]/[C]/[A]/[ICAO].groundnet.xml to be
precise). Setting this property to false will retain the original
behvior.
* For departing aircraft, runway takeoff calculations will be done on the
basis of the performance database. For testing purposes, a performance
estimate for a heavy jet has been added.
notion of a 'displacedThreshold'. Now there's just a real threshold,
displaced or otherwise, and people who care about the paved area can use
'begin' and 'end'. Thanks to John Denker for pointing out the confusion this
leads to. Using 'end' also gets rid of the 'reverseThreshold' name, which was
clearly a bad choice of mine.
This patch changes terrasync so it links against the subversion
library if you have it installed. It supports people who build binary
releases for use by non-developers by removing the runtime external
dependency on having command line svn or rsync available. Since the
patch changes autoconf to detect libsvn, I'd appreciate it if people
who release binaries could verify that the detection scripting works
for their platform.
Developer warning: If you do have libsvn developer libraries
installed, terrasync changes its default option from "-R" to "-S" to
remove the command line dependency. However, Martin has not yet
uploaded world scenery into the subversion repository so it won't be
useful to fly against and you may want to specify "-R" on the command
line in the short term. Or run both.
Me: Update MSVC 7.1 project file. Need svn-win32-1.x.y_dev.zip and svn-win32-1.x.y.zip
located at http://subversion.tigris.org/servlets/ProjectDocumentList?folderID=8100
Convert FGRunway to be heap-based, and inherit FGPositioned. This is a large, ugly change, since FGRunway was essentially a plain struct, with no accessors or abstraction. This change adds various helpers and accessors to FGRunway, but doesn't change many places to use them - that will be a follow up series of patches. It's still a large patch, but outside of FGAirport and FGRunway, mostly mechanical search-and-replace.
An interesting part of this change is that reciprocal runways now exist as independent objects, rather than being created on the fly by the search methods. This simplifies some pieces of code that search for and iterate runways. For users who only want one 'end' of a runway, the new 'isReciprocal' predicate allows them to ignore the 'other' end. Current the only user of this is the 'ground-radar' ATC feature. If we had data on which runways are truly 'single-ended', it would now be trivial to use this in the airport loader to *not* create the reciprocal.
- removes various members from FGRunway which no-one was using
- any of these can be trivially re-instated if and when someone
actually wants to use them - but right now they're simply bloating up
FGRunway, which we have lots of, because it currently includes all the
taxiways in Robin's data.
- that's it.
- Runways are now part of an airport, instead of a separate list
- Runways are no longer represented as a boring struct, but as a class
of their own.
-Improved runway access to unify various runway access methods.
- this exposed a bizarre issue on Mac where dragging in <AGL/agl.h> in
extensions.hxx was pulling in all of Carbon to the global namespace
- very scary. As a result, I now need to explicitly include CoreFoundation
in fg_init.cxx.
- change SG_USING_STD(x) to using std::x
SimGear change. It changes all the SG_xxxx to be the 'real' includes, and gets
rid of many #ifdef SG_HAVE_STD_INCLUDES. As an added bonus, rather than
replacing 'SG_USING_NAMESPACE(std)' with 'using namespace std', I just fixed
the small number of places to use std:: explicitly. So we're no longer polluting
the global namespace with the entire contents of std, in many cases.
There is one more 'mechanical' change to come - getting rid of SG_USING_STD(X),
but I want to keep that separate from everything else. (There's another
mechnical change, replacing <math.h> with <cmath> and so on *everywhere*, but
one step at a time)
PLETE_FUNCTIONAL from SimGear and FlightGear.
As a result, SG_HAVE_STD_INCLUDES is now *always* set, so I will get the boring
fixes for that done, but separately. I'm still auditing the other things in comp
ilers.h - there's a lot that can die now BORLAND is gone.
Various other patches that have been lingering around for a while:
* Moved trafficcontrol.[ch]xx from the Airports directory to ATC, where
it really belongs.
* AI aircraft will request startup clearance, and ground control will
approve.
* Starting AI Aircraft will be pushed back to a predefined holding point
on the ground network, and wait a while before taxiing out to the runway
and the pushback code:
- Traffic manager starts modelload requests immediately upon program
loading
- Only create legs 1 (push back) or five (cruise) of AI traffic.
- AIAircraft's rather obsessive behavior to circle around a waypoint is
largely resolved
- More realistic ground steering for AI aircraft.
From Till:
i started the project at the end of february with a simple idea: move all
3d-model loading to the DatabasePager-thread. my first attempts looked
promising, though they were a little too optimistic (or naive?). the patch
has evolved a lot since.
currently it does the following things:
1. revive SGModelLib, move functions for xml-model-loading there
2. replace all calls to sgLoad3dModel with calls to either
SGModelLib::loadModel() or SGModelLib::loadPagedModel()
almost all models will be loaded by the DatabasePager. the few exceptions are:
your own plane, shared models in scenery, random objects, AIBallistic models.
3. simplify mode-loading functions (avoid passing around fg_root)
4. avoid supurious MatrixTransform nodes in loaded models
5. fix some memory leaks
and "isOnRunway".
- Added initial support for AI controlled pushback operations, making use of the
current editing capabilities of TaxiDraw CVS / New_GUI_CODE. The current
implementation is slightly more computationally intensive than strictly
required, due to the currently inability of taxidraw to link one specific
pushBack point to to a particular startup location. FlightGear now determines
this dynamically, and once we have that functionality in TaxiDraw, the
initialization part of createPushBack() can be further simplified.
- Smoother transition from pushback to taxi. No more skipping of waypoints, and
aircraft wait for two minutes at pushback point.
- The classes FGTaxiNode, FGTaxiSegment, and FGParking, now have copy
constructors, and assignment operators.
- Removed declaration of undefined constructor FGTaxiNode(double, double, int)
- Array boundry checks and cleanup.
- Modified Dijkstra path search algoritm to solve partial problems. Currently
limited to include pushback points and routes only, but can probably be
extended to a more general approach.
- Added initial support for giving certain routes in the network a penalty, in
order to discourage the use of certain routes over others.
