List of known bugs in 1.7.0

[1.7.0 release thread]

I’m trying to gather a list of bugs that need to be fixed for the upcoming release. Please post or link to bugs here so that I can get a complete list.

  • API reference missing [#515440]
  • ODE rejects floating-point values in certain build configurations
  • OSX build shouldn’t use environment.plist
  • Assertion error on showTightBounds() [#518812]
  • GPU is forced into single-precision mode in OpenGL at ATI cards [#514018]
  • str(ode object) does not give anything in the plugin environment
  • p3d tools are not included on Windows, and do not have executable bits on Unix
  • Maya exporters missing on OSX

Lack of vertex colors in Cg shaders can be considered as bug ? Seems it works only for arbvp1.

discourse.panda3d.org/viewtopic.php?t=8387

Maybe also yet another less important thing.

All tk trees has hardcoded color to black/white it should respect system colors.

img705.imageshack.us/i/scr09o.jpg/

OpenCV delivers a plain-white texture image. no errors or warnings.
camera led lights up. the fps change depending on the brightness conditions. so if the camera would send more fps, more are captures. they just show up white for some reason.

ubuntu 9.10 64bit

Bug with egg2bam -ctex -txo: https://discourse.panda3d.org/viewtopic.php?p=52434

There are still problems with OpenAL. When I run the music box sample using OpenAl, there are times when there will be a lot of jittering in the sound and the frame rate will go way down.

If it matters, I have Windows XP.

Maybe also ‘getConfigShowbase not found’ problem for old Linux distros ?

discourse.panda3d.org/viewtopic.php?p=52492

Edit: invalid request [problem solved at least for me]

  • packp3d warns about panda3d.* imports (thanks for fixing, David!)
  • I shouldn’t forget to pick up the OpenSSL 1.0.0 fixes
  • Shadows don’t work for point lights
  • pmerge does not update non-packages in the directory, such as .p3d archives.

Beh. update_frame does get called, there does appear to be valid data. Somehow the ram image gets filled in incorrectly.

David, do you know of any change between 1.6.2 and 1.7.0 that could affect this? OpenCVTexture seems to call do_modify_ram_image() once and then copies the data to:

unsigned char *dest = _ram_images[0]._image.p() + do_get_expected_ram_page_size() * z;

Does this look fine to you, or is it missing a step?

For me enabling OpenAL (without using it even, just enabling in config) causes a significant framedrop - up to 3 times less frames per second.

Additionally, I can’t get PStats to work, it just hangs every time I try to connect to it.

All that on Ubuntu 9.10.

I don’t see anything wrong with that. It does need to call do_modify_ram_image() each frame, though, to set the dirty flags properly.

David

Hm, puzzling. I can’t seem to figure out why it would not work otherwise, while it worked in 1.6.

A divide by zero error occurs if you re-size the Panda window to a height of 0. To reproduce, start the Carousel demo (or any demo) and using the mouse drag the height of the window to be a size 0. Below is a trace dump:

C:\Panda3D-1.7.0\samples\Carousel>Tut-Carousel.py
DirectStart: Starting the game.
Known pipe types:
  wglGraphicsPipe
(all display modules loaded.)
Traceback (most recent call last):
  File "C:\Panda3D-1.7.0\direct\showbase\EventManager.py", line 61, in eventLoop
Task
    self.doEvents()
  File "C:\Panda3D-1.7.0\direct\showbase\EventManager.py", line 55, in doEvents
    processFunc(self.eventQueue.dequeueEvent())
  File "C:\Panda3D-1.7.0\direct\showbase\EventManager.py", line 122, in processE
vent
    messenger.send(eventName, paramList)
  File "C:\Panda3D-1.7.0\direct\showbase\Messenger.py", line 325, in send
    self.__dispatch(acceptorDict, event, sentArgs, foundWatch)
  File "C:\Panda3D-1.7.0\direct\showbase\Messenger.py", line 410, in __dispatch
    method (*(extraArgs + sentArgs))
  File "C:\Panda3D-1.7.0\direct\showbase\ShowBase.py", line 2466, in windowEvent

    self.pixel2d.setScale(2.0 / win.getXSize(), 1.0, 2.0 / win.getYSize())
ZeroDivisionError: float division
:task(error): Exception occurred in PythonTask eventManager
Traceback (most recent call last):
  File "C:\Panda3D-1.7.0\samples\Carousel\Tut-Carousel.py", line 189, in <module
>
    run()
  File "C:\Panda3D-1.7.0\direct\showbase\ShowBase.py", line 2531, in run
    self.taskMgr.run()
  File "C:\Panda3D-1.7.0\direct\task\Task.py", line 496, in run
    self.step()
  File "C:\Panda3D-1.7.0\direct\task\Task.py", line 454, in step
    self.mgr.poll()
  File "C:\Panda3D-1.7.0\direct\showbase\EventManager.py", line 61, in eventLoop
Task
    self.doEvents()
  File "C:\Panda3D-1.7.0\direct\showbase\EventManager.py", line 55, in doEvents
    processFunc(self.eventQueue.dequeueEvent())
  File "C:\Panda3D-1.7.0\direct\showbase\EventManager.py", line 122, in processE
vent
    messenger.send(eventName, paramList)
  File "C:\Panda3D-1.7.0\direct\showbase\Messenger.py", line 325, in send
    self.__dispatch(acceptorDict, event, sentArgs, foundWatch)
  File "C:\Panda3D-1.7.0\direct\showbase\Messenger.py", line 410, in __dispatch
    method (*(extraArgs + sentArgs))
  File "C:\Panda3D-1.7.0\direct\showbase\ShowBase.py", line 2466, in windowEvent

    self.pixel2d.setScale(2.0 / win.getXSize(), 1.0, 2.0 / win.getYSize())
ZeroDivisionError: float division

Oops, that one’s my fault. Just checked in a fix for that.

Don’t forget all the plugin bugs too >_>
I’m stalled for like forever!
I greatly regret installing 3dsmax 2010 now.

I’m checking the snapshot builds repeatingly but it seems you guys are having problems with the windows builds.

What plugin bugs?

Like the exporter from max 2010.
I’m talking about the issues of installing it and when that succeeds it doesn’t even export my model.

I installed max 2010 on my system both 32 and 64bit, downloaded the newest snapshot build (the normal wouldn’t even load), putted the plugins into the 32bit plugin folder.

After that I start the software, it loads but just freezes when I export a model in any method.

=/

I consider this as a bug :
if switching to fullscreen fails, the window properties fullscreen flag is not cleared, so it’s totally misleading.
How do I suppose to know that it fails ?

First time I’ve heard of it. What’s the context–what OS, what display library, and in what way did the fullscreen switch fail?

David