mfranz [Fri, 11 May 2007 18:00:07 +0000 (18:00 +0000)]
- degrade electrical XML SG_ALERT to SG_WARN. The reason why this is output
at all is that this method is meant to be deprecated. But if it doesn't
say that, and there's no generic replacement at all, then it's rather
pointless.
- simplify parameter initialization (The old code copied from Instrumentation/
but has been simplified there in the same manner a few months ago
The new method is less picky about additional nodes ane outputs less
pointless alerts. If there shall really be a check for that, then
this should be a separate function. But it's nowhere else done in fgfs.
mfranz [Thu, 10 May 2007 14:28:17 +0000 (14:28 +0000)]
- don't abort just because of an unexpected "repeat period"! Complain and
recover instead
- it was apparently planned to run the updated loop only every 1000th
frame, but the counter isn't reset, so it runs every frame
- use SG_LOG instead of cerr
- don't add redundant slashes in SGPath::append()
mfranz [Thu, 10 May 2007 12:50:50 +0000 (12:50 +0000)]
- One of the most frequent bug reports is about "terminate called after
throwing an instance of 'sg_io_exception'\nAbandon". And this is caused
by compiling fgfs/sg without exception support (unlikely) or linking
against a libSDL/libglut that was compiled/linked without execption
support. While we can't fix that, we can tell the users who's to blame. :-)
- handle const char * exceptions
mfranz [Wed, 9 May 2007 20:36:43 +0000 (20:36 +0000)]
Maik JUSTUS:
"minor update for the rotor FDM. It results in a more realistic
calculation of the phase shift of rotor and therefor in a little bit
more realistic flight behavior.
(Additionally you can modify the initial position of the rotor and some
(not finished) modifications for the jet ranger rotor)."
mfranz [Mon, 7 May 2007 14:29:40 +0000 (14:29 +0000)]
property_list.cxx: in verbose mode (toggle with Ctrl-'.'-entry) also
show "secret" values, that is: values of nodes with children. These
can be used like all normal (leaf) properties, but their values were
until now only shown in writeProperties() dumps. Also show the
number of attached listeners. Example: foo = 'bar' (string, AU, L3).
dialog.cxx: add warning message for broken <format>s; cosmetics
mfranz [Mon, 7 May 2007 14:06:05 +0000 (14:06 +0000)]
cosmetics: make the reason for the cast obvious. (Now that almost all of
fgfs has to be recompiled (because of a patch to simgear/props/props.hxx)
this is the right time to commit this completely unimportant patch. :-)
mfranz [Sat, 5 May 2007 09:16:59 +0000 (09:16 +0000)]
Declare --props option obsolete and issue a warning message. Rationale:
- props is easy to mix up with the --prop option (for setting properties)
- the name is unspecific and inconsistent: other option names describe
the protocol -- the way to get the properties. How is, for example,
--httpd less about prop(ertie)s?
- two identical options easily confuse people, as can be seen in The
FlightGear Manual, where --telnet and --props were described differently
mfranz [Fri, 4 May 2007 14:58:52 +0000 (14:58 +0000)]
- remove dependence on /sim/number-views: the view manager just uses all
/sim/view[*] that it finds. It's not only unnecessary that view definitions
have subsequent indices, but aircraft are now *requested* to use indices
100++. /sim/view[0] .. /sim/view[99] are reserved for the system. (Not
that we'd ever need that many, This is just a convention, it's nowhere
hard-coded.)
- replace the string operations for property paths by method calls & other
improvements
mfranz [Thu, 3 May 2007 19:00:55 +0000 (19:00 +0000)]
property browser: shift click on the "." entry dumps the contents of that
dir level to the terminal (remember: ctrl-click -> toggle additional info,
and ctrl-click on ".." move to root level)
mfranz [Wed, 2 May 2007 18:49:31 +0000 (18:49 +0000)]
Free the splash texture memory when it's no longer needed. Nowadays splash
textures are mostly 512x512 pixels, so we wasted 786 kB (RGB) or 1MB (RGBA)
during the whole fgfs runtime for mere startup prettification.
mfranz [Tue, 1 May 2007 18:06:48 +0000 (18:06 +0000)]
move the "interpolator" subsystem into the INIT group (same as "nasal"),
because nasal's f_interpolate() may be called in Nasal at times when the
GENERAL subsystem group is being deconstructed; access it by addressing
the group directly, as using globals->get_subsystem() does then not
work any more then; yeah, it's all for a rare border case ... :-)
mfranz [Tue, 1 May 2007 17:03:50 +0000 (17:03 +0000)]
remove util.cxx workaround for listener-call-at-exit-segfault, and make
NasalSys.cxx more robust instead. The reason for the crash was that during
fgfs shutdown destroyed subsystems (GENERAL group) still need Nasal access
(for AI Model destruction listeners), but at that point globals->get_subsystem()
can't even deliver the "nasal" subsystem (INIT group). One way to solve that
problem would have been to replace globals->get_subsystem("nasal") by
globals->get_subsystem_mgr()->get_group(SGSubsystemMgr::INIT)->get_subsystem("nasal"),
but Andy decided to store a pointer to the active "nasal" subsysten in
NasalSys.cxx instead, as the "nasal" subsystem needs to be accessed in
every single Nasal extension function, and multiple "nasal" subsystems are
out of the question, anyway.
mfranz [Tue, 1 May 2007 09:18:11 +0000 (09:18 +0000)]
since the submodel_mgr subsystem does now also hold references to AI models,
we need to explicitly destroy that here, too, so that it has guaranteed
access to the Nasal subsystem. Otherwise we get a segfault on exit. When
the next subsystem needs this special treatement (radar?), we should
introduce a new subsystem group (in addition to INIT and GENERAL)
set viewport in every progress step, so that the splash texture gets rescaled
on window resize in fg/osg (works in fg/plib without that, but can't hurt
there either) (thanks to Tim MOORE)
(hopefully) fix the problem that some people have:
"FATAL: PUI: No Live Interface! Forgot to call puInit ?"
We shouldn't use the splash progress message before guiInit().
Leave the "idle_state" step for now, to keep the similarity
with fg/plib as big as possible. This can be dropped later.
allow rgba splash textures (I don't expect that they'll be used much, but
they don't cost much either, and allow contoured images with colored
backgrounds -- nice for squadron emblems etc.)
Add support for a new file format which where packet type are saved out into
one file per type in a simple binary conglomeration of packets with no headers
or checksumming (this format is intended for local storage only, not to be
transmitted on the fly over a noisy communication pipe.)
Patch to enable atc chatter (and any other arbitrary "message" audio file)
to be played at any specified volume. (Previously these messages were always
played at a hardcoded volume of 1.0)
- save pairs of <file>/<path> for each *-set.xml file in autosave.xml
- loop up file name in the cache at startup. If it's found, check if the
file exists and use it, otherwise scan $FG_ROOT/Aircraft/ as usual,
and create a new cache while doing that. Rebuild cache on FG_ROOT change.
- move fgInitTowerLocationListener() from fgInitPosition() to the init
part of fgIdleFunction() in main.cxx. fgInitPosition() is called again
at every reset, which would every time attach another listener.
setAttribute("archive", 1) sets this attribute to "true", and
getAttribute("archive") returns 1 if the attribute is set, and 0 otherwise.
Allow to query and to set all properties by not specifying an attribute
string: getAttribute() returns all attributes bit coded in an integer,
and setAttribute(attr) sets all attributes. No assumptions may be made
about the meaning of the bits -- they can be changed in future fgfs releases.
The only valid use is to compare or set attribute numbers obtained in the
same fgfs run. This is meant for allowing full copies of property branches.
Also add getAttribute query strings "children" for the number of children,
and "alias".
Csaba "Jester" HALASZ: radar fix & extension
- fix breakage due to former commit (AIManager.cxx, r1.72)
- make AI properties available in AIBase
- add <valid> property for animations/nasal scripts
- support more MP and AI targets
- add target select and altitude display
Csaba HALASZ: don't treat every entry in a scenario file as model entry,
especially those called <description>. This was responsible for the aircraft
at --lon=0 --lat=0.
allow AI models to contain <nasal><load> and <nasal><unload> blocks in
their XML wrapper/animation file. They can access their /ai/models node
via cmdarg() function. Example:
<nasal>
<load>
print("Hi, I'm the Nimitz. My data are under ",
cmdarg().getPath());
</load>
<unload>
...
</unload>
</nasal>
Note, however, that the <unload> block is only called on exit at the moment,
not when the tile is unloaded.
mfranz [Fri, 30 Mar 2007 22:51:52 +0000 (22:51 +0000)]
Vivian MEAZZA:
"""
"Flight plans" which can start at a given time (gmt)
WAITUNTIL tokens which pause the flight plans until a given time (gmt)
Submodels can now be attached to any AI objects (except submodels - it can
be done, but in my experimental code it's too expensive in frame rate atm)
"No-roll" attribute added to Ballistic objects - useful for wakes and the
like
"Random" attribute added to Ballistic objects (adds =- 5% to the Cd) -
useful for smoke, exhausts
If the <trigger> tag is not specified the Ballistic object/s will be
released at start-up (cannot be stopped)
Submodels are not released from AI Objects if the AI Object is more than 15
miles away.
"""
mf: minor code and formatting fixes; submodels.?xx were FUBAR and are thus
astyle formatted;
NOTE that <name> tags END, EOF, WAIT, WAITUNTIL are *depreciated*.
Don't get too used to them. This will have to be moved from the "name"
to regular engries.
andy [Thu, 29 Mar 2007 18:50:28 +0000 (18:50 +0000)]
Sync with Nasal CVS (soon to become Nasal 1.1). Notable new features:
Nasal now supports calls to "subcontexts" and errors can be thrown
across them, leading to complete stack traces when call() is used,
instead of the truncated ones we now see.
Vectors can now be concatenated using the ~ operator that used to work
only for strings.
Better runtime error messages in general due to a fancier
naRuntimeError() implementation
A big data size shrink on 64 bit systems; the size of a naRef dropped
by a factor of two.
"Braceless code blocks" have been added to the parser, so you can
write expressions like "if(a) b();" just like in C. Note that there's
still a parser bug in there that fails when you nest a braced block
within a braceless one.
Character constants that appear in Nasal source code can now be
literal multibyte UTF8 characters (this was always supported for
string literals, but character constants were forced to be a single
byte).
New modules: "bits", "thread", "utf8" and (gulp...) "io". The bits
library might be useful to FlightGear, the utf8 one probably not as
Plib does not support wide character text rendering. The thread
library will work fine for spawning threads to do Nasal stuff, but
obviously contact with the rest of FlightGear must be
hand-synchronized as FlightGear isn't threadsafe. The io library is
no doubt the most useful, as it exposes all the basic stdio.h
facilities; it's also frighteningly dangerous when combined with
networked code...
mfranz [Mon, 26 Mar 2007 15:17:38 +0000 (15:17 +0000)]
layout-props.cxx: hrule/vrule shall be accepted as widgets even if they
have no children (in which case they default to 1px thickness & stretch).
This allows to just write <hrule/> instead of <hrule><dummy/></hrule>.
One can still use <hrule><pref-height>3</pref-height></hrule>, of course.
layout.cxx: drop silly do??Box calls for hrule/vrule (yes, I wrote that :-)
mfranz [Fri, 23 Mar 2007 15:53:58 +0000 (15:53 +0000)]
If text contains "{display|voice}" groups, skip the delimiters and
discard the display part. The curly braces wouldn't be spoken anyway,
and the | would be spoken as "vertical bar", which is completely
useless (which is why it had been disabled in the past already).
mfranz [Sat, 17 Mar 2007 15:58:15 +0000 (15:58 +0000)]
New Nasal command _condition(p) where p is a property node containing
a condition as described in $FG_ROOT/Docs/README.condition. Returns 1 if
condition is true, 0 if false, and nil on error.
mfranz [Sat, 10 Mar 2007 20:46:38 +0000 (20:46 +0000)]
before a day passes with no commits at all, better have some fg-submit stuff :-)
- drop the DEFAULT keyword in .fg-submit configuration files. That was
a silly idea. The default rules are now always appended. One can still
bypass them by ALLOWing or DENYing anything before, for example, by using
DENY *, or ALLOW *.
- fix a typo that broke ~/.fg-submitrc loading (but ~/.fg-submit worked anyway)
- some minor improvments, cleanup and all that
mfranz [Fri, 9 Mar 2007 19:15:54 +0000 (19:15 +0000)]
yeah, yet another patch ... :-]
- fix config file name in $HOME; This didn't match the documentation.
(doesn't cost us anything to check ~/.fg-submit first, and then ~/.fg-submitrc)
- don't use mktemp for the backup files. Some outdated distributions
(Debian) come with a version that mandates six X, which is just too ugly.
Just find the first free slot with sequential number. That isn't thread
safe, but mktemp isn't either, so ... (Should be using "lockfile", but
its availability on CygWin is questionable. And it's not *that* important.)
- some more documentation
- some cleanup, too, of course
mfranz [Fri, 9 Mar 2007 16:22:27 +0000 (16:22 +0000)]
- rules can be defined in optional .fg-submit config files, via commands
ALLOW, DENY, IGNORE, DEFAULT ... see documentation on top
- make fg-upload arguments like the documentation says. (I had accidentally
left $1=$PWD, $2=archive, $3=diff, while it should be $1=archive, $2=diff)
- add -v option (verbose)
mfranz [Thu, 8 Mar 2007 23:16:57 +0000 (23:16 +0000)]
- escape '+' in awk pattern (gawk doesn't need it, but mawk does)
- change the mktemp lines to please Debian's stoneage version ... sigh
- reject archives
mfranz [Thu, 8 Mar 2007 16:45:48 +0000 (16:45 +0000)]
fg-check: speedup, minor fixes, cleanup
fg-submit:
- takes alternative optional basename (used instead of dirname)
- calls optional user defined fg-upload script at the end (example on top)
- minor fixes, different color for changed binary files
- cleanup, improved documentation
mfranz [Wed, 7 Mar 2007 16:26:33 +0000 (16:26 +0000)]
- add fg-check script; Can be used to check source/data before committing or
submitting. Detects various kinds of ugliness, but also reports false
positives. (People aren't supposed to compress texture filer so save
40 bytes. ;-)
- bugfix in fg-submit + some more cleanup and cosmetics
projects / flightgear.git / log