I was having the same trouble with Ubuntu, I figured I was doing something wrong since I just started using Ubuntu two days ago.
The updated version works great. Thanks for the release. You’ve done a lot of hard work and it is very much appreciated.
I’m burning out to test it - looks the new baby so full of stuff I need ages to try every new feature.
By the way I’d a little prc issue after install on ubuntu that actually I’m undecided on how to face it: the install instead to overwrite the old Config.prg in /etc made a new one called Config.prc.dpkg-dist, telling me to investigate it. I found some little changes against the former and I tried to comment both to see which one was used but the engine worked like a charm without them. What should I do? get rid of the former and rename the new, get rid of both or leave everything alone as it is now?
Congrats on a great release!
Is MeshDrawer missing the setPlateSize function? I can’t find anything with dir() that looks like a replacement for it, and there’s nothing in the document (yet, at least).
[edit:] Should have clarified, I only get this problem on Ubuntu.
[edit 2:] Scratch that, I just updated to the official 1.7.0 release on Windows, and now I get the error there too. Wasn’t getting it with the build snapshots.
treeform has updated the MeshDrawer with a new system deprecating the plate system. Ask him for more information about that.
Many congratulations on a successful 1.7 release!
David
Congratulations on the new release!
2 issues I encountered:
-
rdb, the DYLD_LIBRARY_PATH problem seems to be back with the 1.7.0 official release. i’m using the latest version from buildbot and that is fine.
-
panda3d.p3d is not included in the SDK version as far as i can tell (also not with the latest buildbot version).
-
maya2egg*/egg2maya* are not included in 1.7.0. with the latest buildbot version only maya2egg2010/egg2maya2010 is included.
I’ve hit a strange one - I can no longer setup an OdeMass correctly. Whatever functions I call (Including once where I pass the matrix in directly.) the inertia tensor is screwed - never positive definite, and some of the functions leave it with nan’s in. This is obviously fatal for physics - causes ode to fail an assert, and Panda to then exit (And often segfault). Version of Panda I compiled a couple of weeks ago is fine however - its only the new version from the repository that is causing issues (I’m 64 bit Karmic Ubuntu.), and as cvs indicates that no one has touched those files in 7 months I’m guessing some kind of linking issue. Only idea that pops to mind is that the ode header and ode library used for compilation don’t match, and the data structure is now different - I say this because some parameters seem to get through, but in the wrong places.
You can see the problem clearly by running:
from pandac.PandaModules import *
mass = OdeMass()
print mass
mass.setCapsuleTotal(70.0,3,1.0,3.0)
print mass
mass.setParameters(70.0,1.0,2.0,3.0,4.0,5.0,6.0,7.0,8.0,9.0)
print massI also crash when using ODE on Ubuntu, it complains about the mass being zero.
Just a quick follow up:
In regards to et1337, yes, I am getting the same issue - sometimes it complains about the mass being zero rather than the inertia matrix - it depends which setter is called.
I had a look into the idea of the structure getting out of position however, and there are two things I noted when looking through the ode code:
13 months ago the dMass structure was changed - the 4-vector for the centre became a 3-vector, so that could explain the issue, though I think not as its not wholly consistent with the symptoms.
More likely, I note that ode can be compiled to use floats or doubles - a miss-match between the library and the header would cause this problem; alternatively they could match but be compiled with different structure alignment rules, which would probably break only the 64 bit versions (Use floats, align on 64 bits for one, 32 bits for the other.). This last one seems likely to me - it would explain why only a subset of values seem to be getting through, in the wrong positions, why the mass (First entry in the structure) typically is ok, and also why not many people are complaining.
Just to be clear - if you can’t construct an OdeMass you can’t use ode, so until this is fixed ode is unusable, and the thing I was going to release into the showcase sub-forum on Wednesday is in limbo:-(
Oh. I’ll investigate soon - thanks for reporting.
There is no panda3d.p3d. You mean /usr/bin/panda3d ? It’s in the Runtime download.
If you mean the .p3d scripts (packp3d, etc.), they are shipped in /Developer/Tools/Panda3D, with the p3d extension stripped so that they are easily usable from the command line.
Oops, you’re right, I forgot. Sorry.
Ah, you guys are absolutely right about the ODE issue. Coppertop has reported issues before but I didn’t think they were issues with the build. Now I’m convinced.
That could very well be the issue. Thanks for the hint!
Yes, you are right, I was talking about packp3d.p3d. But there is also no packp3d in /Developer/Tools/Panda3D (at least not with my current buildbot install, which is the one from yesterday). There is only a packpanda command line tool. Am I missing something?
You’re right, it’s not in the buildbot release, only in the official release.
I figured out the new MeshDrawer plate system, and I thought I’d post what I found for reference until the manuals are updated.
Basically, instead of passing a plate index to the various functions (billboard(), particle(), etc), you can actually manually specify the UV coordinates in a Vec4, which describes a rectangle, like so: Vec4(x_offset, y_offset, width, height)
The coordinate system has the origin in the lower left hand of the texture, and the upper right corner is (1, 1).
Just in case anyone was wondering. Sorry for the thread hijack.
Hi guys.
I’m having problems with installing the 1.7.0 version.
I’m running on Ubuntu 9.04, 32 bit and by now were using successfully Panda3D 1.6.2
Today I tried the new version, but couldn’t install it at all.
Fisrt the problem was that it was missing libode1, I found the deb with this version, install it, but after this next problem occured: Error: Dependency is not satisfiable: libstdc++6 (>= 4.4.0). My version is 4.3.3
I suppose because I use 9.04, not the newest version 9.10.
I’m not sure if have to upgrade to 9.10 or the problem is somewhere else.
BTW looking the new features I’m just waiting the stable version to start playing with them 
Right, the debs are compiled for karmic and will not work with older versions.
blenderkid and i found ARtoolkit to be completely missing in 1.7.0 . it worked fine for me in 1.6.2 .updating to 1.7.0 i get an error telling me it cant find the ARtoolkit module. downgrading back to 1.6.2 makes it working again. i dont have artoolkit installed manually nor is it available in any repos so i guess the copy which comes with 1.6.2 is the only one.
(karmic 9.10 amd64)
also the entire artoolkit dokumentation is missing in the online-docs.
Oh, you’re right. My apologies. I will fix that soon.
Oh yeah!
Now I have only one last dream, A new stable version, and my game would be totally in a new level of Cool!