Convert the route-manager to use a flight-plan internally, and expose
flightplan, leg and procedure data to Nasal. Move the Level-D parser
into its own file.
Not every xml-autopilot filter needs to run at the update rate
of the FDM. This only makes sense for the autpilot filters. Those,
who update instrument needles or environment properties for example
may run at frame rate.
This patch (thanks to ThorstenB, who is the author) forces those
filters defined in /sim/systems/autopilot into the FDM loop and those
defined in /sim/systems/property-rule into the default loop. They
show up as xml-autopilot and xml-proprules in the performance monitor
- initialize uninitialized properties
- use prefix instead of postfix increments
- reduce visibility of variables
- use empty() instead of size() == 0 for vector and string
- pass string by reference, not by value
- Add update-interval-secs to the entire autopilot
- cache min/max values in InputValue
- a little more relaxed "equals zero" checking in the NoiseSpikeFilter
- add new <update-interval-secs> for autopilot components.
This does what it says.
- add new method to PeriodicalValue to normalize symmetric around zero
- move the DigitalFilterImplementation out of the header file
- refactor NoiseSpikeFilter: spare some cpu-cycles and respect periodical
output
Whenever resolving a (relative) path to an absolute path with
'resolve_maybe_aircraft_path', check if the result is empty and report original
(relative) path as missing. Otherwise no or a meaningless message is
issued ("File '' not found.").
The tag <initialize-to> can be used to control the value
of the output when the component is first enabled. This
controls initialization of the output property and the current
value for internal computation
Valid values are
<initialize-to>input</initialize-to>
set the output-property to the input value
<initialize-to>output</initialize-to>
set the output-property to the output value
<initialize-to>none</initialize-to>
ignore input and output value
Signature of Predictor::configure must match AnalogComponent::configure,
otherwise the inherited method isn't overridden.
=> predictor couldn't be configured.
=> speed predictor rules in "generic-autopilot-helper.xml" weren't working.
The autopilot has grown beyond being just an autopilot. It's
components, filters and controllers could be used to set property
values based on complex rules and driven from other property values.
That's why I chose to give it the name "property-rule".
This patch allows to define an arbitrary number of property rules under
a/sim/property-rule[n]/path
using the same syntax as autopilot configurations.
Add helper to look for an aircraft branch path in multiple candidate locations.
Update the main subsystems to use the helper, and hence load from any aircraft dir.
- Respect the global inverted flag in the get_output() method
- Check the clock state before processing the static R/S inputs
- Emit debug output only on a state change
Split the former single-file-implementation in xmlauto.?xx into multiple
files and use some OO techniques. Started with documentation to be used
with doxygen.
- reuse code
- undeclared inputs default to false where applicable
- remove unused tInput
- JK, D and T FF now inherit from RS with dominant R input, so they all
have a R and a S input to preset the output
- fix the nonfunctional <inverted> element of the logic element
Old calculation was
output = input * gain
and is now
output = (input-reference) * gain
Note: the PI(D) controller use (reference-input) which effectively reverses the sign. Our notation was picked for backwards compatibility to not break the myriads of <filter> elements currently existing.
All the rest is some code cleanup.
"logic filters" use well known conditions to drive output properties. Example for bax = baz & (foo | bar).
<logic>
<name>my first logic element</name>
<input>
<or>
<property>foo</property>
<property>bar</property>
</or>
<property>baz</property>
</input>
<output>bax</output>
</logic>
Supports various new editing features, including dragging to re-order, and
+/- keys to adjust the target altitude for a waypoint. Also displays some
additional information, and will display *even* more once I land airways/
SID/STAR support.
- default to an invalid altitude in routes, instead of cruise altitude (temporarily)
- only set an altitude on the autopilot, if valid
- only add departure airport/runway to the route, if not airborne
- allow multiple <autopilot> elements within an aircraft. All autopilot live in an individual FGXMLAutopilot subsystem which run within a subsystem group now.
* When the nav-radio is slaved, calculated radial/target-hdg-deg
(needed by some autopilot logic)
* Handle editing (including deletion) of route waypoints correctly,
including deleting the active waypoint
* Add a signal to the route manager when the last wpt is reached, and
use it in the GPS to revert to OBS mode.
* Change the altitude handling to use the specified cruise altitude
* Fix a bug where autopilot/locks/altitude was treated as a boolean
<abs>true</abs>
for input elements.
If set to true, the input value is filtered thru fabs() function.
Defaults to false if absent, so there is no impact for existing configurations
- a refactoring of the route manager to reduce indentation and duplication
- created a helper method to generate the ETA strings
- created a helper to update target_altitude and altitude_set when the active wp0 changes
- used early returns to make update() easier to follow
- removed spurious includes in both header and source file
Attached patch updates the route-manager to use FGPositioned to search
for waypoints, instead of a manual airport / fix / navaid search. This
is good because it's now using the 'strictly closest' match, rather
than arbitrarily picking a distant fix over a nearby navaid. In my
case, the TLA VOR is significant to several EGPH procedures, but also
happens to be the ident of a fix a long, long way away.
Also updates the FGPositioned class to stop using Point3D, partly
because it's deprecated and partly because I had misunderstood the
interface and was using it wrong. For now, all FGPositioned distance
checks use SGGeodesy::inverse, which is accurate but inefficient. Once
FGPositioned queries are used for something on a hot path, I'll
probably store the cartesian position as well as the geodetic, to make
these checks fast.
Here's part 2 - converting FGFix (the simplest one) to be both heap-based and inherit FGPositioned. One minor benefit from this is replacing some dangerous code in FGFixList which used to return the address of an iterator member ('&it->second'). To keep the diff a sensible size, I'm not updating the callers to use the richer FGPositioned types - i.e replacing separate lat/lon handling with SGGeod. I will make those cleanups, but in future patches.
- 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
