Bugs. Segfault etc...
Greetings. breve looks great, but it is not working.
I just downloaded the latest version of the IDE for Linux, and installed it to a sub-subfolder in my home directory. I opened the readme to get started. It told me to export classpath, seemed to go fine. Then it said I can start the demo by writing "./breve ./demos/Gatherers.tz". There is no "breve" executable in the breve folder, but I tried with the other executables (breveIDE), and the project opened. THen, when I hit "play", it stopped with the following errors (from the log):
[Gatherers.tz]
an error occurred executing the internal function "newCube": invalid size for new sphere (side <= 0.0)
Begin stack trace:
breve engine tack trace:
0) Gatherers (0x8f3a028) init (line 92 of "Control.tz")
1) Cube (0x8f80528) init-with (line 222 of "Shape.tz")
End stack trace:
... error in file "Shape.tz" at line 225
Afterwards, when I try to open files from within the program, it just crashes with the following shell prints:
PaHost_OpenStream: could not open /dev/dsp for O_WRONLY
PaHost_OpenStream: ERROR - result = -10000
PaHost_OpenStream: could not open /dev/dsp for O_WRONLY
PaHost_OpenStream: ERROR - result = -10000
Segmentation fault (core dumped)
And the breveIDE_ex gave instant Segmentation fault upon execution one time.
Have I mistaken anything?
non English locale?
Are you by chance using a non-English locale? Another user reported a similar problem running with a French locale, and I think this might be related.
It's possible that using "env -i ./bin/breve ..." might help get past this problem.
- jon
Hmm..
Yes. I am using Norwegian regional settings.
I now also downloaded the command-line version and tried to run both programs with "env -i ./bin/breve". It changed nothing. But I have discovered that running Python simulations works, although laggy (I have an old computer). The command line version loaded the steve file, but hung at the first frame. Then it crashed when I right-clicked inside the window.
please try this new build
Hi there,
I'd like you to try a new build which should fix the problems you've encountered.
Sure enough, it looks like the locale settings were interfering with the steve code parsing and causing problems. There's also a command-line version which fixes the quit you described.
- jon
Yes
If it were not for the fact that breve is the most interesting program I have encountered, I would have given up by now.
So I look forward to trying the new build when I get back to my Ubuntu computer.
I tried the screensaver on a Norwegian Windows XP computer. It shows 24 tries (although not random), and then stops proceeding during the 25th. Number 25 just continues like "forever". If the screensaver is interrupted and then started again, it crashes after few seconds on the 25th try. The next generation never shows. Creatures.tz in the IDE works, however.
I am testing the Windows version of the IDE on this computer now. It works, for some examples of both .tz and .py, with no hacking. But the program crashes when I try to open one file after another. I have to quit in order to reset something. And I cannot open files from subfolders, most of the time. Unknown errors here and there. Not usable. I hope the new version fixes all these bugs, also on WinXP.
Betas updated again
I apologize for the continued problems you've been having and I thank you very much for reporting them here.
As a result of your reports, I have updated the betas again, this time for all platforms and have also included a new build of the screensaver for Windows.
I believe that now all of the problems you've reported should be fixed.
- jon
Beta for Windows now tested a bit.
The new screensaver works fine.
IDE:
DLA.tz, Gatherers.tz, DLA.py, Gatherers.py, PatchLife.py, PatchLife3D.py, Aggressor.tz, BraitenbergTemplate.tz, Coward.py/.tz, GrayScott.py, HyperCycle.py seem to work fine.
PatchLife.tz crashes the program.
PatchLife3D.tz:
warning: unused class variable "archiveList" in class Matrix
Aggressor.py does not attack the red balls.
BraitenbergTemplate.tz:
warning: unused class variable "leftSensor" in class myBraitenbergControl
warning: unused class variable "leftWheel" in class myBraitenbergControl
warning: unused class variable "rightSensor" in class myBraitenbergControl
warning: unused class variable "rightWheel" in class myBraitenbergControl
GrayScott.tz:
warning: unused class variable "archiveList" in class Matrix
Grayscott3D.py (aborted):
[GrayScott3D.py]
Traceback (most recent call last):
File "", line 87, in ?
File "", line 18, in __init__
File "", line 42, in init
File "lib/classes\breve\PatchGrid.py", line 290, in initWith
return self.initAt( x, y, z, center, pSize, patchclass )
File "lib/classes\breve\PatchGrid.py", line 241, in initAt
self.gridPointer = breve.breveInternalFunctionFinder.patchGridNew( self, gridCenter, self.patchSize, x, y, z )
File "lib/classes\breve\__init__.py", line 139, in __call__
return breveInternal.callInternalFunction( self.function, args )
RuntimeError: invalid type for argument 0 of internal function "patchGridNew" (got "double", expected "vector" )
GrayScott3D.tz:
warning: unused class variable "archiveList" in class Matrix
warning: unused local variable "k" in method init of class GS
warning: unused local variable "j" in method init of class GS
Hypercycle.tz:
warning: unused class variable "archiveList" in class Matrix
warning: unused class variable "inflow" in class GS
warning: unused class variable "temp" in class GS
warning: unused class variable "tempreact" in class GS
warning: unused local variable "sum" in method iterate of class GS
That was all before the "Get started" menu.
Creatures.tz: I loaded the old "save". It runs fine, but about twice or thrice each generation one of the creatures is aborted with a black screen with the following text at the bottom: Distance travelled: -1#IND00
It may well be because it is an old and buggy save from the older version of the program, but I thought it best to mention it.
SuperWalker.py made a few tries before crashing with this error message: AttributeError: 'MonkeyGenome' object has no attribute 'getValue'
I will hopefully test the rest soon too. :-)
Segfault Issue on Mac osX 10.5
Good Day!
I have created a multi-agent swarm that results in the following thread crash in the ODE function:
dInternalStepIsland_x2(dxWorld*, dxBody* const*, int, dxJoint* const*, int, double) +7158
My simulation starts off with about 600 agents and ramps up to around 2000+ agents after about 2500 world iterations. In addition, the simulation is set up to register a very large number of collisions between the agents in the system. Anywhere between 2000-3000 iterations, about 80% of the time, I get a segfault crash in the above function. :-(
After doing some searching on the net, I have come across a few others using ODE that have had the same problem and suspect that the problem relates to the default stack size (8mb) on the Mac OSx platform. From what I have read, ODE uses more and more of the stack space the more collisions that occur (thoughts from anybody on this?).
I think the fix for this would be to add:
-stack_size 4000000
to the Other Linker Flags field of the Build Styles pane in xcode, to increase it to the maximum of 64mb.
I did download the project builder source code for Breve, but after several hours of trying, have had no luck in getting it to compile. :-(
Would it be possible to get a Mac version of Breve built configured to use the maximum stack space?
Thanks Much!!!
Titan