Re: [Mokomaze] Two bugreports and a feature-request (was: Re: [Mokomaze 0.5.0] is released)
Marcel-2 So sid's ODE package is kinda useless for us because we need single precision, correct? Yes, I think so. build our own ODE 0.11 with single precision... I've already packed it. But it would be great to have properly-packed version in pkg-fso repository. The Digital Pioneer if you could just do everything in ints, that would be best. Is there any way to pull that off? There is no integer-only version of ODE. -- View this message in context: http://n2.nabble.com/-Mokomaze-0.5.0--is-released-tp2754592p2800244.html Sent from the Openmoko Community mailing list archive at Nabble.com. ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: [Mokomaze] Two bugreports and a feature-request (was: Re: [Mokomaze 0.5.0] is released)
neovento Could you provide a link to the version that works? I tried the lenny package, but no success. Aww. I've finally got my Hackable:1 working and I've tested the .deb package. After installing libode0.11 from unstable it asserts (because of double precision). After installing libode0.9 from stable - the game trying to find libode0.11 (yes, the executable is compiled for using single-precision libode = 0.11). So, I've packed a .deb of an up-to-date ode0.11 with single precision. Now it works and the game runs faster than on my SHR testing. NEW: I've released a new revision of the game - 0.5.1. Now .deb depends from my libode, not libode0debian1. One more feature introduced: anti-aliasing [1] Check the homepage [2]. [1] http://mokomaze.projects.openmoko.org/images/anti-aliasing.png [2] http://mokomaze.projects.openmoko.org/ -- View this message in context: http://n2.nabble.com/-Mokomaze-0.5.0--is-released-tp2754592p2800527.html Sent from the Openmoko Community mailing list archive at Nabble.com. ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: [Mokomaze] Two bugreports and a feature-request (was: Re: [Mokomaze 0.5.0] is released)
Am Dienstag, 5. Mai 2009 19:03:26 schrieb ANT: build our own ODE 0.11 with single precision... I've already packed it. But it would be great to have properly-packed version in pkg-fso repository. I'd love to do that, but it will take some time since I'm just beginning with packaging and won't have time to do it before next tuesday. My plan looks like making a proper, working package for mokomaze before I try something more complex to get into the process. (Hadn't intended to tell you until I had finished... ;D) I'll see how I get along. -- Marcel ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: [Mokomaze] Two bugreports and a feature-request (was: Re: [Mokomaze 0.5.0] is released)
Moin. Am Montag, 4. Mai 2009 07:05:56 schrieb ANT: neovento wrote: Can you fix this soon? That would be really great. :) I am trying to prevent ode's (debian sid) asserts, but no luck. Code of the game is correct. I've surfed the web and all recommends to recompile ode with --enable-release if that assert appears. I've found it was ./configure'd with --enable-double-precision and without --disable-debug, so the problem is in one of this flags. Anyway if some coding trick will be found to avoid the problem, Double Precision will drop the performance. Currently, install libode0debian1 from stable and have a fun. double precision might kill performance on the neo, that's true. But in general it would rather cause internal ODE errors - the asserts should only be affected by disable-debug. So sid's ODE package is kinda useless for us because we need single precision, correct? -- Marcel ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: [Mokomaze] Two bugreports and a feature-request (was: Re: [Mokomaze 0.5.0] is released)
I was under the impression that the FR has no FPU at all, so single precision may be better than double, but if you could just do everything in ints, that would be best. Is there any way to pull that off? ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: [Mokomaze] Two bugreports and a feature-request (was: Re: [Mokomaze 0.5.0] is released)
Am Montag, 4. Mai 2009 17:36:01 schrieb The Digital Pioneer: I was under the impression that the FR has no FPU at all, so single precision may be better than double, but if you could just do everything in ints, that would be best. Is there any way to pull that off? Aren't integers single precision? If yes, we just have to stick to Debian Lenny's ODE package or build our own ODE 0.11 with single precision... (Which would imho be the better solution since ODE improves continually [does that word exist?]) -- Marcel ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: [Mokomaze] Two bugreports and a feature-request (was: Re: [Mokomaze 0.5.0] is released)
continuously On Mon, May 4, 2009 at 5:42 PM, Marcel tan...@googlemail.com wrote: Am Montag, 4. Mai 2009 17:36:01 schrieb The Digital Pioneer: I was under the impression that the FR has no FPU at all, so single precision may be better than double, but if you could just do everything in ints, that would be best. Is there any way to pull that off? Aren't integers single precision? If yes, we just have to stick to Debian Lenny's ODE package or build our own ODE 0.11 with single precision... (Which would imho be the better solution since ODE improves continually [does that word exist?]) -- Marcel ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: [Mokomaze] Two bugreports and a feature-request (was: Re: [Mokomaze 0.5.0] is released)
I believe both are words. It is my understanding that you can have single precision floating point numbers (float) or double precision (double), and integer values are simply ints. I could be wrong though. It did happen once before. ;) ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: [Mokomaze] Two bugreports and a feature-request
The Digital Pioneer wrote: I believe both are words. It is my understanding that you can have single precision floating point numbers (float) or double precision (double), and integer values are simply ints. I could be wrong though. It did happen once before. ;) Single precision floating point numbers use 32bit and double precision floating point use 64bit according to IEEE754 standard for floating point numbers. So double precision needs calculation with longer numbers and needs more cpu power with no hardware fpu unit present. Ciao, Rainer ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: [Mokomaze] Two bugreports and a feature-request (was: Re: [Mokomaze 0.5.0] is released)
Paul Fertser Thank you for this great app, it's really nice and cool and fun and addictive. It is nice to hear that is works normally on Debian. And unfortunately it'll need some fixing because libode1 from Debian sid asserts with your app. Thanks for report. I'll investigane with what 'configure' parameters was compiled libode1 (from sid) and try to prevent asserts. But first of all I need to reinstall by broken debian. Try to make it resolution-independent. It now segfaults in QVGA mode. I assume it as a long term plan: there are many physical values and graphical objects to be scaled down when computing or drawing. I guess we all would like to have a special cheat: if you hold FR upside down, the ball shouldn't fall into a hole :) 1 of 3 people that saw how the ball fell down say that :) I will think something up. -- View this message in context: http://n2.nabble.com/-Mokomaze-0.5.0--is-released-tp2754592p2776798.html Sent from the Openmoko Community mailing list archive at Nabble.com. ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: [Mokomaze] Two bugreports and a feature-request (was: Re: [Mokomaze 0.5.0] is released)
neovento wrote: Can you fix this soon? That would be really great. :) I am trying to prevent ode's (debian sid) asserts, but no luck. Code of the game is correct. I've surfed the web and all recommends to recompile ode with --enable-release if that assert appears. I've found it was ./configure'd with --enable-double-precision and without --disable-debug, so the problem is in one of this flags. Anyway if some coding trick will be found to avoid the problem, Double Precision will drop the performance. Currently, install libode0debian1 from stable and have a fun. -- View this message in context: http://n2.nabble.com/-Mokomaze-0.5.0--is-released-tp2754592p2785659.html Sent from the Openmoko Community mailing list archive at Nabble.com. ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
[Mokomaze] Two bugreports and a feature-request (was: Re: [Mokomaze 0.5.0] is released)
Hey ANT! Thank you for this great app, it's really nice and cool and fun and addictive. Marcel tan...@googlemail.com writes: Could you consider building against libode1 (ODE 0.11, debian sid) instead of libode0debian1 (ODE 0.9, debian lenny/squeeze)? Freerunner-Debian is a Debian sid/experimental so it would be nice not having to pull libode0debian1 from squeeze... Yeah, i second this. And unfortunately it'll need some fixing because libode1 from Debian sid asserts with your app. That's the first bugreport. # mokomaze File_loader: savegame file not found File_loader: 14 game levels parsed Accelerometer: starting auto-detection of device type Accelerometer: Neo Freerunner detected Accelerometer: current threshold value is 0 Accelerometer: threshold is not required to be changed. ODE Message 2: mass must be 0 in dMassCheck() File mass.cpp Line 49 ODE Message 2: mass must be 0 in dMassCheck() File mass.cpp Line 49 ODE INTERNAL ERROR 1: assertion dMassCheck(mass) failed in dBodySetMass() [ode.cpp] Aborted The second: Try to make it resolution-independent. It now segfaults in QVGA mode. How to reproduce: xrandr -s 240x320 (for X.org it's enough, for broken Xglamo you'll need the second command: echo qvga-normal /sys/bus/spi/devices/spi2.0/state Now you'll see a QVGA screen (but with xglamo default configuration the TS still reports 640x480 coordinates, therefore TS is unusable; xorg doesn't have that problem), try to start mokomaze via ssh, it segfaults: Program received signal SIGSEGV, Segmentation fault. [Switching to Thread 0x40022230 (LWP 8828)] 0xaf48 in CreateSurface () Current language: auto; currently asm (gdb) bt full #0 0xaf48 in CreateSurface () No locals. #1 0xc89c in render_window () No locals. #2 0xa144 in main () No locals. (gdb) Ok, and one little feature-request: I guess we all would like to have a special cheat: if you hold FR upside down, the ball shouldn't fall into a hole :) Great thanks and Happy hacking! -- Be free, use free (http://www.gnu.org/philosophy/free-sw.html) software! mailto:fercer...@gmail.com ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: [Mokomaze] Two bugreports and a feature-request (was: Re: [Mokomaze 0.5.0] is released)
Am Samstag, 2. Mai 2009 20:50:58 schrieben Sie: Ok, and one little feature-request: I guess we all would like to have a special cheat: if you hold FR upside down, the ball shouldn't fall into a hole :) LOL. So the game mustn't end when the ball drops so that we can recover it... Maybe even by moving the FR up sharply to throw the ball a little upward... :D -- Marcel ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community