|
|
|
Attached ~/Library/Logs/Vivox/siplog.txt.
According to the system requirements page, 32-bit versions of Linux are supported. If you are using a 64 bit Linux, then you need to make sure the 32-bit compatibility environment is installed.
http://secondlife.com/corporate/sysreqs.php Having said that, I'm not sure the voice product supports 64bit Linux as it stands. Couldn't the company make a 64bit version of their library so someone could create a 64bit SLVoice? This is rediculous, 64bit starting to dominate and SL is stuck in the past with 32bit only versions.
On Linux, the SLVoice executable parses /proc/cpuinfo at startup and seems to have trouble with empty entries like this:
"power management:" This causes the std::out_of_range-exception. It's probably a fault in the parser of libvivoxsdk.so and not really related to 64 bit systems. I use kernel 2.6.22, not sure if this applies to all kernel versions. I have no idea if Linden devs can do anything about it. If not, one possibility would be to patch the kernel to remove the empty entry. I am experiencing exactly the same issue on my new laptop (core2duo 64bit).
Although voice is working fine me on my AMD64 desktop. @Carjay: I just compared the output of my /proc/cpuinfo on both machines and on my desktop it says "power management: ts fid vid ttp", whereas it is empty on my laptop where it is not working. Did you already try to patch the kernel and see if the problem persists? If it doesn't, could you please attach the patch here or make it public somewhere? Would be nice if this could be tracked down... Another possibility would be to install a 32bit Gentoo on my laptop which I don't really wanna do until now. I just tried Carjay's assumption with the empty entries in /proc/cpuinfo.
Result: SLVoice works now! (Although my soundcard has no hardware mixing (intel_hda), I only have voice and no other sound yet, but its a good start) Steps to repoduce: 1. Open /usr/src/linux/arch/x86/kernel/setup_64.c @Linden Labs: Please send this issue to Vivox so they can fix their /proc/cpuinfo parser, just tell them it crashes on empty entries like: "bla :" where the entry doesn't have any value. Shouldn't be a serious problem for them to fix. Nice troubleshooting, I've passed on the info.
I posted an in-depth report on this issue back in March that pinpointed the exact cause of the problem and detailed a temporary solution for users; one that does not require the kernel to be modified. That report can be found here: http://jira.secondlife.com/browse/VWR-5708
A representative from Vivox had informed me that my information was passed on to the developers, hopefully they'll make this more of a priority now Tofu is on board. Reopening with info from bug report
Boy, I hope a kernel patch is NOT being considered as a long-term solution. My OS should be considered authoritative, NOT the SL client.
No... that would be pretty silly!
We have a SDK fix in-hand. The fix will very likely land by the viewer 1.23 release.
Here's a workaround: cd into the "lib" subdirectory of the place where you've unpacked Second Life, and do:
cp /proc/cpuinfo ./ chmod 755 cpuinfo echo >> cpuinfo cp -p libvivoxsdk.so libvivoxsdk.so-orig sed -i 's/\/proc\/cpuinfo/lib\/cpuinfo\c@\c@/g' libvivoxsdk.so Note that this is only necessary through 1.22.RC7. After that, the SLim first look version of the vivox binary will be merged in, and that does not have this problem. (However, I have seen some bad voice quality problems with the Vivox binary in the SLim first look tree As Prospero mentions above, this fix is now in the 1.22 Release Candidate RC8. Using RC8 should resolve this problem without any further workaround.
Hello to you,
I have a very big problem I am no longer able to connect to Second Life since I reinstall Windows Vista Business SP1 French. Can I have a fairly large machine and then power to see the power of my computer go my site website: http://aide-et-conseil-de-windows.net/Mon_Ordinateur_Personnel.html "Logoman Rexen"
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
|
Bug
Resolved
Normal
Terminate called after throwing an instance of 'std::out_of_range' 
)
The following is printed to standard output after enabling voice in the RC viewers; note the out_of_range exception thrown:
2008-02-17T03:55:13Z INFO: LLVoiceClient::setState: entering state stateStart
2008-02-17T03:55:13Z INFO: LLVoiceClient::setState: entering state stateDaemonLaunched
2008-02-17T03:55:14Z INFO: LLVoiceClient::setState: entering state stateConnecting
2008-02-17T03:55:14Z INFO: LLVoiceClient::setState: entering state stateIdle
2008-02-17T03:55:14Z INFO: LLVoiceClient::setState: entering state stateConnectorStart
2008-02-17T03:55:14Z INFO: LLVoiceClient::clearCaptureDevices: called
2008-02-17T03:55:14Z INFO: LLVoiceClient::addCaptureDevice: Advanced Linux Sound Architecture (ALSA)
2008-02-17T03:55:14Z INFO: LLVoiceClient::clearRenderDevices: called
2008-02-17T03:55:14Z INFO: LLVoiceClient::addRenderDevice: Advanced Linux Sound Architecture (ALSA)
2008-02-17T03:55:14Z INFO: LLVoiceClient::setState: entering state stateConnectorStarting
terminate called after throwing an instance of 'std::out_of_range'
what(): basic_string::substr