These are chat archives for highfidelity/hifi

30th
Jun 2014
AlericInglewood
@AlericInglewood
Jun 30 2014 00:27
I created a pull request for my Job - should I change the state of the job now?
AlericInglewood
@AlericInglewood
Jun 30 2014 12:23
What does 'svo' stand for?
[DEBUG] [2014-06-30 03:52:58 +0000] [4621:4619] [particle-server] persistFilename=resources/particles.svo
Leonardo Murillo
@murillodigital
Jun 30 2014 12:48
@AlericInglewood svo == sparse voxel octree
AlericInglewood
@AlericInglewood
Jun 30 2014 12:49
Thanks :). We just started our own domain server... but everything is entirely empty. How can one create the first voxel?
damn, interface asserted... [2014-06-30T14:51:04] ASSERT: "alloc >= 0 && extra >= 0" in file tools/qbytearray.cpp, line 72
AlericInglewood
@AlericInglewood
Jun 30 2014 13:00
Is there a manual on how the Tools --> Metavoxel Editor works?
AlericInglewood
@AlericInglewood
Jun 30 2014 13:06
Hmm Tools --> Script Editor is not doing anything, except closing the chat window. What is it supposed to do?
Judas
@judasshuffle
Jun 30 2014 13:41
Mesh is only appearing as black hole (maybe a back ball) tried on kevin's server as the sandbox only has 3 servers up
Thijs Wenker
@thoys
Jun 30 2014 13:47
@aleri
Use the 0 key to place the initial voxel using editvoxel.js active
AlericInglewood
@AlericInglewood
Jun 30 2014 13:49
Thanks! :)
Currently looking into controlling avatar / cam with mouse and alt-control-shift keys, starting with lookWithMouse.js
It would be nice if there was documentation about this :P .. Is there any?
Clearly there is a 'MyAvatar' object available in .js scripts... but shouldn't that be documented somewhere? What can you do with it?
AlericInglewood
@AlericInglewood
Jun 30 2014 15:23
#14 QCoreApplication::sendSpontaneousEvent good
AlericInglewood
@AlericInglewood
Jun 30 2014 15:51
#7 QCoreApplication::notifyInternal good
#3 QGLWidget::event bad
AlericInglewood
@AlericInglewood
Jun 30 2014 16:05
#5 QApplicationPrivate::notify_helper good
#4 GLCanvas::event bad
Chris Collins
@cozza13
Jun 30 2014 16:13
@judasshuffle did that just happen?
Judas
@judasshuffle
Jun 30 2014 16:13
last 5 mins asfar as i can tell
AlericInglewood
@AlericInglewood
Jun 30 2014 16:16
That one is BIG!
Chris Collins
@cozza13
Jun 30 2014 16:16
any idea on who/how it was created?
AlericInglewood
@AlericInglewood
Jun 30 2014 16:17
I'm in my own domain and only pressing control or Alt keys :P
Chris Collins
@cozza13
Jun 30 2014 16:18
cool
AlericInglewood
@AlericInglewood
Jun 30 2014 16:19
The ALT key is the only key that alternatingly is not delivered to Application::keyPressEvent and therefore not to scripts. Any idea where to look for why that is the case? I can't find it with gdb.. too much going on with events :/
Judas
@judasshuffle
Jun 30 2014 16:20
is there anyway u can get an svo of the sandbox a back up if there is one could maybe put on server, with a view to salvaging kryssies build?
Chris Collins
@cozza13
Jun 30 2014 16:20
we have a backup
AlericInglewood
@AlericInglewood
Jun 30 2014 16:20
What is pressing ALT supposed to do? It seems interface switches to another mode - a menu thing? @cozza13
Judas
@judasshuffle
Jun 30 2014 16:20
was thinking coudl get kev to put it on his server
Chris Collins
@cozza13
Jun 30 2014 16:26
one sec @AlericInglewood just sorting out sandbox
AlericInglewood
@AlericInglewood
Jun 30 2014 16:26
Of course! :)
AlericInglewood
@AlericInglewood
Jun 30 2014 20:26
Duh.. I add a virtual function to QObject, so I could test inside Qt5 if some object is some hifi object (for debugging)- but that took so long that I regretted not using ccache. So, I set up ccache with a size of 8GB - then inverted my patch and compiled Qt5 AGAIN - too late realizing I was overwriting my previous one-hour wait :/. So, after this I'll have to compile it with patch AGAIN.. and I bet that the 8GB isn't going to be enough to even contain two copies of Qt in the cache :(, so what I'm waiting for now is probably also for nothing sigh
AlericInglewood
@AlericInglewood
Jun 30 2014 20:34
Ugh, it's already at 3 GB ... hopefully it will stay under 4 GB :/
Yay - 3.7 GB :)
AlericInglewood
@AlericInglewood
Jun 30 2014 22:01
I ran into an assert in Qt (because I compiled it with debug support), and tracked it down to being a bug in Qt :/. This sucks, I guess we can't fix bugs in Qt - should I try to work around the bug in hifi?
Clément Brisset
@Atlante45
Jun 30 2014 22:02
What kind of bug is it?
AlericInglewood
@AlericInglewood
Jun 30 2014 22:03
@Atlante45 When we try to make an ssl connection and there is no error whatsoever, then we reach this code in Qt:
QSslSocketPrivate::_q_resumeImplementation (this=0xf93200) at ssl/qsslsocket.cpp:2336
2336 q->setErrorString(sslErrors.first().errorString());
But calling first() on an empty container is undefined.
If you look at http://codepad.org/F12egK3L
ignoreErrorsList is empty, ignoreAllSslErrors is false. Hence verifyErrorsHaveBeenIgnored() returns false.
That is a logic error when there are no errors at all.
Sure, no errors were ignored - but you want that function to return true then because when all errors are ignored then there is no difference not having any error at all.
A work around could be to make sure that ignoreErrorsList is not empty - just add some error to it that never happens.
Clément Brisset
@Atlante45
Jun 30 2014 22:07
Wow
That’s weird, I though we were alredy using SSL, does this mean we always get errors?
AlericInglewood
@AlericInglewood
Jun 30 2014 22:09
I don't know - everything worked fine for me until my last recompile of Qt5 and hifi - I don't even understand why I get this assert now and not before. But, I started to look at the code of Qt itself first to understand what the assert means and my conclusion is that this is a bug... I might be wrong though...
Clément Brisset
@Atlante45
Jun 30 2014 22:12
What version of Qt did you recompile?
AlericInglewood
@AlericInglewood
Jun 30 2014 22:13
The same one I used before :P .. 5.2.1
This happens immediately after this line in the logs:
[2014-06-30T23:49:53] Faceshift: "Connection refused"
That isn't an SSL error, so that might explain it...
Clément Brisset
@Atlante45
Jun 30 2014 22:16
But that line isn’t an error, it’s just debug.
let me look at the code
AlericInglewood
@AlericInglewood
Jun 30 2014 22:17
void Faceshift::noteError(QAbstractSocket::SocketError error) {
if (!_tcpRetryCount) {
// Only spam log with fail to connect the first time, so that we can keep waiting for server
qDebug() << "Faceshift: " << _tcpSocket.errorString();
}
it's a socket error though. The ssl code above is reached as a result of a "posted event" - so it could be delayed. If the QSslSocketPrivate::_q_resumeImplementation() is a result of the above failure (ie, we failed but don't want to give up) then this code could be checking if it makes sense to give up, or if the ssl error can be ignored. However, connection refused isn't an ssl error .. so that's where it goes wrong then imho.
Clément Brisset
@Atlante45
Jun 30 2014 22:20
Are you using faceshift?
AlericInglewood
@AlericInglewood
Jun 30 2014 22:21
no - I have no special hardware at all.
Actually, I'm just guessing here :P .. I have no idea what faceshift is.
Chris Collins
@cozza13
Jun 30 2014 22:22
@AlericInglewood is this on your debian machine?
Clément Brisset
@Atlante45
Jun 30 2014 22:22
it’s a facial tracking software
AlericInglewood
@AlericInglewood
Jun 30 2014 22:22
@cozza13 yes .. @Atlante45 then I guessed right.
Clément Brisset
@Atlante45
Jun 30 2014 22:22
We use it to track and reproduce facial expression.
AlericInglewood
@AlericInglewood
Jun 30 2014 22:23
I just remembered Sean mentioning that hifi is using faceshift when we were looking at Mixamo :P
Apart from the fact that I didn't get this assert before, while nothing changed that should have made a difference, I'm fairly sure that this is a bug in Qt and the right thing to do is ignore all ssl errors - meaning, we need to fill ignoreErrorsList with a bogus error by calling QSslSocket::ignoreSslErrors
AlericInglewood
@AlericInglewood
Jun 30 2014 22:29
Hmm the only errors to ignore that you can add are one of these: http://codepad.org/1RriyMmQ
NoError would be nice :)
Clément Brisset
@Atlante45
Jun 30 2014 22:31
It might be the behaviour expected actually
I bet a NoError error should be in that list if there is no error.
AlericInglewood
@AlericInglewood
Jun 30 2014 22:32
The behavior is calling code with undefined behavior @Atlante45
You can't call first() on an empty container.
Oh, no I don't think so. This is a list of real errors that you want to ignore.
Wanting to ignore NoError is nonsense, since it isn't even an error.
Clément Brisset
@Atlante45
Jun 30 2014 22:33
Ohhhhh ok, that’s not the actual error list
AlericInglewood
@AlericInglewood
Jun 30 2014 22:33
So, if it allows us to add that then that would be the perfect work around.
No, that enum is used to be able to pass an error code around while there is no error (as opposed to duplicating code / functions).
The containers will just be empty when there is no error, I think.
Clément Brisset
@Atlante45
Jun 30 2014 22:35
Ok, I just wonder how that didn’t show up earlier.
AlericInglewood
@AlericInglewood
Jun 30 2014 22:35
Me too
I suppose I can try to go back to the old binary - by reverting whatever it was that I did - .. and then debug this code to see what it does then...
Clément Brisset
@Atlante45
Jun 30 2014 22:39
Can you try cleaning you setting files first?
I wonder if it could come from there.
AlericInglewood
@AlericInglewood
Jun 30 2014 22:47
This faceShift connection doesn't even appear to be an SSL connection? Is it?
Clément Brisset
@Atlante45
Jun 30 2014 22:48
No but there is an option to use faceshift or not, it might be checked.
AlericInglewood
@AlericInglewood
Jun 30 2014 22:48
Where can I turn that off?
Clément Brisset
@Atlante45
Jun 30 2014 22:49
in the ini file.
What OS are you on?
AlericInglewood
@AlericInglewood
Jun 30 2014 22:49
debian
Clément Brisset
@Atlante45
Jun 30 2014 22:49
let me check
ter
AlericInglewood
@AlericInglewood
Jun 30 2014 22:50
sean:~/.config/High Fidelity>ls
Interface.ini
Clément Brisset
@Atlante45
Jun 30 2014 22:50
Check there: ~/.config/High\ Fidelity/Interface.ini
Eheh
AlericInglewood @AlericInglewood is fast :)
AlericInglewood
@AlericInglewood
Jun 30 2014 22:52
Ok, it was true - now false. "Connection refused" line disappeared - assert still there. So, it was not relelated to faceShift (connection).
Clément Brisset
@Atlante45
Jun 30 2014 22:53
Arf
AlericInglewood
@AlericInglewood
Jun 30 2014 23:05
sigh - it takes like 2 minutes to start interface in gdb - it's sooo slow reading the debug info :mad:
And with a break point set it runs at 1/100th of the speed :\
Drake Arconis
@CmdrCupcake
Jun 30 2014 23:23
Does hifi need SDL1.x or 2.x on linux?
Clément Brisset
@Atlante45
Jun 30 2014 23:24
You don’t need SDL
We are using it for some hardware we are working with that is not yet available to the public
So it should compile fine even without it.
Drake Arconis
@CmdrCupcake
Jun 30 2014 23:25
oh okay. trying to get it building on archlinux :P
Clément Brisset
@Atlante45
Jun 30 2014 23:25
@CmdrCupcake did you encounter any issue because of it?
Ahah, nice.
Drake Arconis
@CmdrCupcake
Jun 30 2014 23:26
It mostly just works besides a wall of cmake warnings due to cmake 3.0 deprecations
Awh. Segfault on start of interface :(
AlericInglewood
@AlericInglewood
Jun 30 2014 23:33
@CmdrCupcake You're stalking me :P
Sean Devin
@SeanDevin
Jun 30 2014 23:33
lol @AlericInglewood
Drake Arconis
@CmdrCupcake
Jun 30 2014 23:33
\o/ silly segfault was caused by GCC 4.9.x being horridly broken