1
0
Fork 0
Commit graph

167 commits

Author SHA1 Message Date
andy
3362ea2447 A bug was discovered on IRC where an errant script was setting a nil
value on a property.  This becomes a NaN when converted to a numeric
value, which then percolated into the C++ world where it ultimately
caused a crash in YASim's turbulence code.  While converting nil to
NaN isn't *strictly* wrong, it's dangerous for this reason.  Toss a
Nasal exception instead.  Hopefully this won't break too much
preexisting code.
2006-10-17 19:58:33 +00:00
andy
0c109dabbb More warnings, similar issues. 2006-08-08 18:32:17 +00:00
andy
503cc54a93 Rework the naModLock()/naModUnlock() semantics to be more
maintainable.  The rules are simple (extension functions are called
*with* the lock, which must be dropped before calling naCall(), which
grabs it) but the tracking of when the lock was held was getting a
little confused.  Keep a "nasal call depth" count in the subsystem to
figure out whether we are making a sub-call and thus hold the lock.
2006-07-19 19:46:53 +00:00
mfranz
4914844c31 ... but then again: it's only a rumor that listeners on tied properties
don't work. And like most rumors, it's untrue. So just output a debug
message.
2006-07-16 11:48:22 +00:00
mfranz
1841bd7343 - complain about attempts to attach a listener to a tied node
- do *not* complain about a listener writing to its own node, but don't
  enter recursion either
2006-07-16 11:30:33 +00:00
mfranz
e48967cb1d fix another crash on exit by finally converting the rest of unguarded
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.
2006-06-11 10:21:10 +00:00
mfranz
00f3d1db80 - remove listener in ~FGNasalListener(), not FGNasalSys::removelistener
- delete all listeners in ~FGNasalSys()
2006-06-10 22:21:22 +00:00
mfranz
e0b1d12a60 don't abort fgfs only because a nasal script called a property function
with an invalid path, as in getprop("/sim/model/737") or x.getNode("f:1").
Forward sg's error message to the Nasal runtime error function instead, so
you get something like:

  Nasal runtime error: name must begin with alpha or '_'
    at /home/m/fgfs/Base.local/Nasal/props.nas, line 30

Unfortunately, the location points to the line where the ghost wrapper
sits, rather than the offending script line.
2006-05-23 18:55:38 +00:00
mfranz
2ae061c9c3 explain createModule()'s optional SGPropertyNode argument 2006-04-27 16:40:04 +00:00
mfranz
52306b9093 add optional SGPropertyNode* argument to NasalSys::createModule. This is
used in dialog.cxx to allow XML dialogs access to their own prop tree via
Nasal's cmdarg(). That way dialogs can generate dynamic content, such as
list entries.
2006-04-27 15:56:51 +00:00
mfranz
7845f33c74 a bool isn't safe enough here, but a mutex shouldn't be necessary 2006-03-20 11:01:16 +00:00
mfranz
4c5e96f730 - disallow listener re-entry to prevent crash/runaway process
- move FGNasalListener function bodies to NasalSys.cxx (has become too big)
2006-03-20 07:13:10 +00:00
andy
8ca3e913ad Add a (commented out) line enabling the "bits" library for testing. 2006-03-15 19:43:09 +00:00
andy
a9ea522388 Add the I/O library, but comment it out until we figure out how to do this securely 2006-03-15 18:10:29 +00:00
mfranz
5f339a4e83 close Nasal/ dir after reading *.nas files 2006-03-13 19:47:22 +00:00
mfranz
31d3fd0209 add alert message 2006-03-09 17:32:24 +00:00
mfranz
b9d3c3ccb6 check if the "nasal" subsystem is still there, and only call <unload> then.
This isn't the case if the model is destroyed on fgfs exit. To make <unload>
work under these circumstances, one would have to reorder subsystem removal,
but this doesn't seem overly useful, so we'll do it when we really need it.
2006-03-09 13:30:28 +00:00
mfranz
83b64b0140 remove leftover #include; changed comment 2006-03-09 11:05:32 +00:00
mfranz
2796f3f4a5 NasalSys.[ch]xx:
implement FGNasalModelData class for execution of XML <load> and <unload>
  scripts. modelLoaded() is called by the model loader, and the destructor
  on branch removal.

modelmgr.cxx:
tilemgr.cxx:
tileentry.[ch]xx:
  make scenery and custom objects run their Nasal scripts on loading
  and unloading. Let OBJECT_STATIC object not be cached.
2006-03-09 09:04:03 +00:00
mfranz
6867462210 crashfix: handleCommand() is called from other modules (input callbacks,
fgcommands from C++ or GUI), so we need a separate context here. (See
also solution in FGNasalListener::valueChanged)
2006-03-08 16:06:32 +00:00
mfranz
77de0c8e09 add deleteModule() method 2006-03-08 10:35:20 +00:00
mfranz
dcffb9d335 let removelistener() return the number of remaining Nasal listeners. (We
need to return nil for errors and something else for success, so we can
as well return something (remotely) useful.)
2006-03-02 10:41:48 +00:00
mfranz
55cde5f160 - assign unique id to listeners
- let setlistener return id
- add removelistener(id) command
2006-02-28 14:55:37 +00:00
fredb
36e4045810 Add missing include files needed by the new math code under windows 2006-02-18 13:58:09 +00:00
mfranz
034e39dba5 screenPrint() is obsolete. Use screen.log.write() for the same purpose,
or write to the color properties in /sim/screen/. If this Nasal/GUI
implementation turns out to be too slow, we'll write a generic OpenGL/plib
version simliar to the ATCdisplay code.  (OK'ed by Andy and Stuart)
2006-02-13 20:08:04 +00:00
mfranz
3a2148828e screenPrint() is obsolete. Use screen.log.write() for the same purpose,
or write to the color properties in /sim/screen/. If this Nasal/GUI
implementation turns out to be too slow, we'll write a generic OpenGL/plib
version simliar to the ATCdisplay code.  (OK'ed by Andy and Stuart)
2006-02-13 19:46:03 +00:00
mfranz
d1fa3f9c7e store garbage collector id (again) in the listener class, and free the
handler on destruction. There's no explicit Nasal command for removing
a listener, but deleting the node should fully remove them.
2006-02-04 16:37:25 +00:00
mfranz
f9bf9fd073 fix for deadlock problem (as suggested by Andy) 2006-02-02 18:03:07 +00:00
mfranz
9efe35b86a support optional setlistener() arg to fire callback function initially 2006-01-30 12:29:58 +00:00
mfranz
0b136fa66b getChild(): add optional "create" flag (like in the c++ version) 2006-01-27 20:43:40 +00:00
mfranz
0d48631e3a cosmetics: subc isn't really a sub-context here (it was where I copied it
from)  -->  s/subc/c/
2006-01-27 19:59:25 +00:00
mfranz
32968d81f3 we need a way to specify other contexts than _context; idea & OK by Andy 2006-01-27 19:51:25 +00:00
mfranz
f0996fa47d create a new context for listener functions to avoid context corruption
(caused a crash when a timer triggered a listener) ... Idea and OK by Andy
2006-01-26 00:18:27 +00:00
curt
691f096c68 Add an srand() function to nasal (hooked into sg_srandom_time()).
This helps avoid the problem of always getting the same random number
sequence every time you run a script.
2006-01-09 15:29:24 +00:00
curt
71d0001f3d Add a directory listing command to nasal. Given a directory name, return
the file list as an array.
2006-01-09 03:48:14 +00:00
ehofman
0e16419f0d Stuart Buchanan:
- Provide a Nasal interface to display simple text messages on the screen
  like the ATC display. In fact, I copied the code from the ATCDisplay.cxx
  and simply shifted it further down the screen.


Erik:

TODO: Integrate the two pieces of code.
2006-01-06 09:50:58 +00:00
mfranz
96c0a7bbc1 add _setlistener() function, so that slow property polling loops can be
replaced with efficient listener callbacks. One use is the new FPS display.
This is reviewed and OK'ed by Andy, relatively trivial and separated from
the rest of Nasal, so problems are quite unlikely and confined to users of
this function.

The callback is executed whenever the property is written to -- even if
the value didn't change. The triggering Node is available via cmdarg().

Examples:  _setlistener("/sim/crashed", func {print("haha!")});
           _setlistener("/foo/bar", func { print(cmdarg().getPath() ~ " changed")})
2005-12-16 19:11:03 +00:00
andy
14c3bc5932 Architectural fix allowing the "tip" popups (FOV, view name, etc...)
to pop themselves down while the simulator is paused.

The problem was with the "real time" queue in the event manager,
causing the third argument of Nasal's settimer() (a flag for "sim
time") to be ignored.  Inverts the default sense of the argument, as
there are lots of uses of settimer() in the current code, almost none
of which want to use real time.

Note this fix introduces a header file incompatibility in SimGear --
be sure to update.
2005-11-09 20:34:46 +00:00
mfranz
75877ebf43 ... and don't keep. This causes nasty, mysterious bugs and has no advantages. 2005-10-23 16:13:53 +00:00
mfranz
cad193a1c4 add support for removeChildren(): takes one name as optional argument
and removes all children with this name; if no name is given, removes
all children  (OK'ed by Andy)
2005-10-23 16:10:32 +00:00
andy
0f704b0a6e Changes to support the new naCall API in SimGear. NOTE: these changes
are *not* backwards compatible, and require current SimGear to
compile.
2005-09-20 21:09:52 +00:00
mfranz
d977c07908 - output property path in error messages 2005-06-20 18:53:00 +00:00
mfranz
ece07db538 - rename initModule() to createModule() (as suggested by Andy)
- make it public
- enable handleCommand() to execute a binding (or other nasal code defined
  in a property system subtree) in a particular namespace (-> "module" child)
2005-06-19 17:09:03 +00:00
andy
568d53875e Make readScriptFile() public, and rename it to loadModule(), so that external
code can have access to the facility for loading modules from files.

Also includes a cast-int-to-pointer fix that causes a warning on 64 bit
systems.
2005-06-14 19:57:24 +00:00
andy
3a6bc316ed Same bug, different place. Don't treat a length of zero as "please
strlen() it", because real files and properties can have lengths of
zero.
2005-04-26 20:57:06 +00:00
andy
239077599c Don't try to strlen() the input buffer if the length is zero. That
was probably my idea of a feature, but if the input buffer actually
does has a length of zero (as Melchior discovered for the case of a
zero-length .nas file) then there will be no null.
2005-04-26 17:18:00 +00:00
andy
142854d3d1 Upgrade to Nasal 1.0 test candidate 2005-04-18 19:49:13 +00:00
curt
ecfaa79b9e Permit a single nasal module to reference any number of files which are all
loaded into that single nasal name space.
2005-03-16 21:36:55 +00:00
ehofman
def25f39d5 gcc 4.0 fixes. 2004-12-18 11:59:33 +00:00
andy
d43d10046b Melchior caught a missing return value that is now an error on newer gcc versions 2004-11-15 18:15:33 +00:00
andy
e590737451 Fix a bug where the interpreter could read a garbage buffer from the stack. 2004-05-15 21:40:58 +00:00
curt
2acdd02879 Clean up various compiler warnings that have crept into the code. This
by no means get's them all, but it's a start.
2004-04-01 15:27:53 +00:00
andy
3ddfd77781 Forgot to handle "unspecified" properties in getprop(). Treat them
as strings.
2004-03-24 19:06:54 +00:00
andy
b3ce2c3cfe Failed timers should log errors rather than silently eating them. 2004-03-24 18:37:58 +00:00
andy
2166dd70a8 Forgot to implment the Node.getChild(name, index) form of the function 2003-12-29 22:52:21 +00:00
andy
7edecf980e Workaround for code generation bug in gcc-2.95.3 2003-12-16 05:39:04 +00:00
andy
0b3159ec0c Add a new command, "dialog-new", allowing external code (e.g. Nasal
scripts) to create dialogs at runtime.  Augment "dialog-close" to take
a name argument, allowing code other than PUI callbacks to close
dialogs.

The changes to the GUI directory to enable this are actually minor,
basically amounting to using SGPropertyNode_ptr reference counting
(the GUI subsystem no longer "controls" the dialog property trees, so
it can't delete them).
2003-12-08 02:05:10 +00:00
andy
eb3f52c340 Turn the __timers garbage collector cache into a more generic "gcsave"
interface, and use it to cache FGNasalScript objects returned from
a new parseScript() method.

Added a rand() function.

Added an interpolate() function interface to the new SGInterpolator
subsystem.
2003-12-05 01:54:39 +00:00
ehofman
79e48829da Add Nasal Vs. 1.5 2003-12-01 14:35:49 +00:00
ehofman
2be85c1b1b gcc 2.95 and 3.0 requite handleTimer to be public 2003-11-27 14:34:37 +00:00
curt
d1360d26d0 Remove old PSL support and put our eggs into the Nasal basket. 2003-11-25 21:16:13 +00:00
curt
d3181fb437 Andy Ross:
A set of changes to impliment Nasal support in FlightGear.  The nasal
interpreter is now part of SimGear.
2003-11-25 21:08:36 +00:00
ehofman
980012e168 Move FGEventMgr and FGSubsystemMgr over to SimGear, add SGEventMgr to FlightGear's globals structre and some small code cleanups 2003-09-24 17:20:55 +00:00
curt
80e98bdc57 Compiler squawked because a typed function had no return, create a dummy value
to return.
2003-08-11 19:33:49 +00:00
david
04f1dc4837 Make all PSL support conditional, because it's not included in plib
1.6.  The macro is HAVE_PLIB_PSL.
2003-01-17 12:41:47 +00:00
david
af4f1894c9 Added a run_inline method that automatically wraps the script in a
main() function.  It's not hooked to a command, yet.
2003-01-16 18:04:24 +00:00
david
9bbe8a78b6 Added simple proof-of-concept support for running PSL scripts from
input bindings.  They will work only with the latest CVS; otherwise,
./configure will disable them.  There is a new command, 'script',
which takes a single argument, also called 'script', containing PSL
code (currently PSL requires a main() function).

Erik Hofman has written some more elaborate code for triggering PSL
code from drop-down menus and scheduling events; I will look at
integrating that next.
2003-01-11 16:30:32 +00:00