the next airport or airport with METAR station, but about any type of
airport
- as a side effect this change makes it also 30 to 50% faster :-)
In the long run this linear search shall be replaced with a spatial
algorithm (like octree), which will be a much bigger performance gain.
Turn OPENSCENEGRAPH_MAJOR_VERSION, OPENSCENEGRAPH_MINOR_VERSION and
OPENSCENEGRAPH_PATCH_VERSION into a single number for comparisons in the
preprocessor.
Anders said:
With Stuart's help I've looked closer at this and I think I've tracked
down the cause of the problem:
At least on my computer the sort() call on line 234 in
Environment/environment_ctrl.cxx sorts the vector entries by memory
address instead of altitude, i.e. the custom comparison predicate is not
used. This causes the tables of environment conditions to be reordered
into a wrong order at some weather updates, depending, basically,
on where the memory allocator places the objects. (Btw. why are they are
freshly allocated for each update?)
"interpolates METAR changes over time as follows:
1) Wind changes are interpolated as vectors, so the change is much more
natural than before - no-longer will your aircraft be suddenly shoved to
one side.
2) Visibility is interpolated as an X-value, which looks pretty neat.
3) Cloud heights and thicknesses are interpolated if they are close to the
aircraft. Cloud textures are not."
"bind the visibility-property of the SGSky object to a property
named /envirionment/effective-visibility-m. This property is needed to decide
if the aircraft is within clouds or not. I use this property also for the
structural icing code."
SGPropertyNode to guarded ones. This is also done for JSBSim/JSBSim.hxx,
for which JSB had given explicit permission a while ago. I postponed that
back then, but now is the time.
mf:
pthread_cancel doesn't seem to work correctly on all supported platforms.
It apparently causes SGBlockingQueue::pop() to correctly leave the thread on
the cancellation point pthread_wait(), but the SGGuard() destructor isn't
called, so the queue mutex remains locked. This triggered an assert() on
pthread_join(). This patch uses an empty ICAO request as signal for the
thread to terminate itself.
- remove cleanup handler that unlocks unused mutex
- make the result_queue a locking queue (blocking is only done in the
pop() anyway, and this isn't used in the thread at all)
- don't disallow thread cancelling in the request_queue's pop(), which
is the only cancellation point in this thread
The prototype of update_metar_properties does not match overridden func.
trafficmgr: iterators below begin() and after end().
tower.cxx : iterator incrementing beyond end().
metar fetcher. Effectively this caused the metar thread and the main
thread to both attempt to fetch weather data. This could lead to long pauses
when the main thread decided to fetch the weather, and introduced a race
condition that could cause a segfault/crash.
Investigating this issue, I discovered that even longer ago, someone confused
#defines and #ifdef symbols with C/C++ variables. If I #define XYZ 0 it is
defined so #ifdef XYZ is true, not false like a variable. Our thread
detection made this mistake and there were follow up patches to work around
it.
So I fixed the configure script (ahhh, reading the autoconf manual is highly
recommended excercise for people editing the configure.ac file.) I also
discovered that we were hardwiring with_threads=yes with no way via configure
options to disable threads from the build so I fixed that.
Then I patched up the #ifdef's scattered through the code to match the
configure script changes, oh and by the way, I stumbled upon a past typo
that led to the race condition in the metar fetching thread and fixed that.
too old (wrong system time or broken proxy): stops after 10 stale reports
were fetched in a row. This should simply stop further fetching, but due
to a bug in the threading system(?) it does currently lead to abortion,
just like any other exception in the fetcher.
I tried to make sure accessor functions which return by reference act
on const objects. also replaced some iterators with const_iterator
and a few return/pass by reference that were missed the first time
around.
* Use "const string&" rather than "string" in function calls when appropriate.
* Use "const Point3D&" instead of "Pint3D" in function calls when appropriate.
* Improved course calculation in calc_gc_course_dist()
* Safer thread handling code.
Vassilii Khachaturov:
Dont use "const Point3D&" for return types unless you're absolutely sure.
Erik Hofman:
* Use SGD_(2)PI(_[24]) as defined in simgear/constants.h rather than
calculating it by hand every time.
CygWin/gcc-3.4.4 updates.
I replaced my cygwin compiler with 3.4.4, did a make clean of plib, simgear,
and flightgear, then did a make install of all three. With the included changes,
everything builds fine, and runs fine.
Attached is a patch to the airport data storage that I would like committed
after review if acceptable. Currently the storage of airports mapped by ID
is by locally created objects - about 12 Meg or so created on the stack if
I am not mistaken. I've changed this to creating the airports on the heap,
and storing pointers to them - see FGAirportList.add(...) in
src/Airports/simple.cxx. I believe that this is probably better practice,
and it's certainly cured some strange problems I was seeing when accessing
the airport data with some gps unit code. Changes resulting from this have
cascaded through a few files which access the data - 11 files are modified
in all. Melchior and Durk - you might want to test this and shout if there
are problems since the metar and traffic code are probably the biggest
users of the airport data. I've also added a fuzzy search function that
returns the next matching airport code in ASCII sequence in order to
support gps units that have autocompletion of partially entered codes.
More generally, the simple airport class seems to have grown a lot with the
fairly recent addition of the parking, runway preference and schedule time
code. It is no longer just an encapsulation of the global airport data
file, and has grown to 552 bytes in size when unpopulated (about 1/2 a K!).
My personal opinion is that we should look to just store the basic data in
apt.dat for all global airports in a simple airport class, plus globally
needed data (metar available?), and then have the traffic, AI and ATC
subsystems create more advanced airports for themselves as needed in the
area of interest. Once a significant number of airports worldwide have
ground networks and parking defined, it will be impractical and unnecessary
to store them all in memory. That's just a thought for the future though.
- replay.cxx :
corrected a bug, now reinitialize the recording data when replay is
deactivated
- fgclouds.cxx :
cloud layers and weather condition are saved when choosing a weather scenario,
added a new scenario 'none' so we can switch back to standard flightgear
weather
- navradio.cxx :
force a search() on init to initialize some variables, preventing a nearly
infinite loop when delta-time == 0 on the first update()
- electrical.cxx :
uninitialized variable in apply_load() for FG_EXTERNAL supplier
- panel.cxx, panelnode.cxx :
added a property "depth-test" for 2.5D panels so that they update the depth
buffer and are no more visible from the outside of the aircraft when the
aircraft uses textures without an alpha channel
- panel.cxx :
moved the computation of the instruments diffuse color outside the
texturelayer code since this is constant during a frame, this is a big speedup
for 2D panels
I did some profiling of the code and found a few interessant things. Some corrections are obvious like the one in the multiplayer code, the fps is no more divided by 2 or 3 when another plane is on screen.
Other things like collision detection and computation of agl can not really be optimized. I changed a few things in hitlist.cxx but this only give a very low increase of fps. The groundcache eats a lot of cpu but I think that the real way to do it is to use a real collision system like OPCODE or something like that.
And I added an option to disable the recording of replay data. It takes more cpu than we can think.
Changes
=======
- panel.cxx :
moved the computation of the instruments diffuse color outside the texturelayer code
since this is constant during a frame, this is a big speedup for 2D panels ;
- hitlist.cxx :
changed the computation of the intersection between ray and triangle, optimized
the sphere culling by using a normalized direction vector. This can give a
35% speedup on the computation of elevation in some situations ;
- renderer.cxx, acmodel.cxx :
call ssgDrawAndCull with plane scene graph in external or internal view,
calling ssgDrawAndCull with the root scene graph was drawing other players plane
a second time in multiplayer mode ;
- mplayer.cxx :
removed the calls to ssgFlatten and ssgStripify because it was degenerating models,
causing a massive drop in frame rate ;
- replay.cxx :
added an option to disable the recording of the flight
- fgclouds.cxx :
changed the path of cloudlayer properties to match preferences.xml ;
set the altitude of clouds from scenarios to a more correct value if metar is not enabled ;
Changes
=======
- correct the transparency probleme when old 3d clouds were enabled
(rendering context with an alpha channel)
- changed rain cone orientation, it can now be viewed from helicopter or chase
view (still not tower view)
- clouds are a bit more yellow/red at dawn/dusk
- weather data is now correctly propagated to the interpolator, this correct
visibility, wind, etc
- the 'metar' weather scenario now immedialty reuse the real metar data
- real metar no more overwrite custom weather scenario
This is another update for the cloud code, a lot of lines but this time I have started to add the doxygen doc.
Misc
====
- corrected a bug when RTT is not available, the current rendering context was
altered
- if RTT is not available then 3d clouds are not drawn at all
- impostors lighting is now recomputed when the sun changes position
- distant objects are no more seen in front of clouds
- blending of distant clouds is a bit better now
- litle optimization of code (uses a less cpu time)
- use layer wind speed and direction (no more hardcoded wind)
- fov is no more hardcoded
Changes
=======
- clouds (cu only) are dissipating/reforming (experimental)
- compute a turbulence factor that depends on surrounding clouds and type of
clouds (experimental)
- clouds shapes are defined in cloudlayers.xml
- type of clouds present in a layer is also defined in cloudlayers.xml
- cloud layers are generated from metar and other misc. data (in progress)
- added a rain effect around the viewer (enabled in the rendering dialog and
when the metar property says so)
- added a lightning effect (enabled in the rendering dialog) : cb clouds spawn
new lightnings
- added a dialog to select from different weather source : metar/property,
a 'fair weather' environment and a 'thunderstorm' environment.
Changes
=======
- corrected some strange behavior when playing with the render dialog options
- the density slider is now working : if you are fps limited and still want to see clouds in
the distance you should play with that
- added the choice for texture resolution, its more comprehensible now (before it was
wrongly allways choosing 64x64 textures)
- changed the initial texture size : you now have 64 texture of 64x64, this uses 1Mo of
texture memory (before it was 20 texture of 256x256, that took more memory and there was
not enought impostors)
- sun vector is now right so the lighting is a bit better
- removed useless sort and light computations for impostors, this should save a lot of cpu
- blending of distant cloud is more accurate now
- clouds are now positioned correctly, they don't try to escape you anymore
- no more red/white boxes around cloud
- textures are now filtered (no more big pixels)
known bugs
==========
- distant objects are seen in front of clouds
- new and updated sources for the new volumetric clouds
- 2 new textures for the clouds
- an update to the render dialog to enable/disable and change a few parameters
for the new clouds
ssgSetNearFar(). This by default creates a symmetric view frustum which is
typically what an application wants.
However, to get control of the view frustum in order to build support for
asymmetric view frustums, we need to wrap these calls with a bit of our own
logic.
This set of changes wraps all calls to ssgSetFOV() and ssgSetNearFar() with
FGRenderer methods.
I also standardized how the FGRenderer class is handled in globals.[ch]xx.
This led to some cascading changes in a variety of source files.
As I was working my way through the changes, I fixed a few warnings along
the way.
I just heard from John Wojnaroski that you and he are going to work on getting
a flightgear demo machine up for the linux expo thursday and Friday. John
indicated that he would very much like to get a CVS version with the new
traffic code up and running before the expo.
The following patches to SimGear & FlightGear ...
- create an FGMetar abstraction layer, whose purpose is:
* provide defaults for unset values
* interpolate/randomize data (GREATER_THAN)
* derive additional values (time, age, snow cover)
* consider minimum identifier (CAVOK, mil. color codes)
- add rain/hail/snow/snowcover support on the METAR side
- add max age of METAR data handling (currently set to
- add support for an external METAR cache proxy server
- add CAVOK handling
- set missing year/month in regular METAR messages
- fix a small bug in metar.cxx (wrong return value)
a single apt.dat.gz file which is in the native X-Plane format.
To do this I wrote a front end loader than builds the airport and runway
list. Some of the changes I needed to make had a cascading effect, so there
are minor naming changes scattered throughout the code.
