Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Repo info
Activity
Sebastian Jennen
@zebastian
...because appveyor tried to run the test on its own, failed, and did not export the artifact :D. I deactivated the test again in latest commit, should be available in 30-60 minutes here: https://ci.appveyor.com/project/zebastian/mandelbulber2-s84yl
Sebastian Jennen
@zebastian
oh no, msvc (again) wont compile.
(ClCompile target) -> C:\projects\mandelbulber2-s84yl\mandelbulber2\src\audio_track.cpp(207): error C2131: expression did not evaluate to a constant [C:\projects\mandelbulber2-s84yl\mandelbulber2.vcxproj] C:\projects\mandelbulber2-s84yl\mandelbulber2\src\audio_track.cpp(210): error C3863: array type 'double [fftSize*]' is not assignable [C:\projects\mandelbulber2-s84yl\mandelbulber2.vcxproj] C:\projects\mandelbulber2-s84yl\mandelbulber2\src\audio_track.cpp(211): error C3863: array type 'double [fftSize*]' is not assignable [C:\projects\mandelbulber2-s84yl\mandelbulber2.vcxproj] C:\projects\mandelbulber2-s84yl\mandelbulber2\src\audio_track.cpp(222): error C2440: 'initializing': cannot convert from 'double [fftSize*]' to 'double' [C:\projects\mandelbulber2-s84yl\mandelbulber2.vcxproj] C:\projects\mandelbulber2-s84yl\mandelbulber2\src\audio_track.cpp(223): error C2440: 'initializing': cannot convert from 'double [fftSize*]' to 'double' [C:\projects\mandelbulber2-s84yl\mandelbulber2.vcxproj] 770 Warning(s) 5 Error(s)
thats inside the calculateFFT in the new cAudioTrack:
this line fails to compile double fftData[fftSize * 2];
since fftSize is not recognizable as a constant expression
msvc needs it as a constant, or the array has to be dynamically allocated with new
http://stackoverflow.com/questions/33423502/expression-did-not-evaluate-to-a-constant-c
maybe best to always use the constant cAudioFFTdata::fftSize?
Krzysztof Marczak
@buddhi1980
I have just changed that to cAudioFFTdata::fftSize. Let see if the program will be compiled
Sebastian Jennen
@zebastian
thanks
Krzysztof Marczak
@buddhi1980
mandelbulber --test doesn't work at all under win10
but the aplication starts properly with gui and NetRender can be used with this compilation
however program is no so stable. Hangs sometimes when material is edited (probably problems with threads) and crashed one time after rendering
it doesn't happen when I use mingw version
Sebastian Jennen
@zebastian
hmm, do you know some way to get a stack trace?
Krzysztof Marczak
@buddhi1980
I'm going to install MSVS Express. Then I will get debugging capabilities
Krzysztof Marczak
@buddhi1980
it will take 14 gigabytes on disk :confused:
Sebastian Jennen
@zebastian
Quality has a size :P
Sebastian Jennen
@zebastian
hey rob, do you know whats going on with appveyor? last build has been a few days ago, all newer commits are only in queue. maybe general outages of the service or something about repository config?
Sebastian Jennen
@zebastian
just made a test with raspberry compilation over night
compiles and runs like a charm :)
Krzysztof Marczak
@buddhi1980
Great! Low long it takes to compile it?
Sebastian Jennen
@zebastian
it was 9 and a half hours. It was no clean make, but used 20 precompiled classes from last build in 2016.
So overall it would take probably 11 hours for a full compile on a raspberry model b single core, thats only half the time of a full render of the example aexion fract file, which estimated to ~24 hours :)
Sebastian Jennen
@zebastian
can you rerun coverity scan?
Krzysztof Marczak
@buddhi1980
No, I cannot. There is a bug in Qt ot coverity-build tool (already reported). I need to install old Ubuntu 12.04 (LTS) to have it running
Krzysztof Marczak
@buddhi1980
I have bring CoverityScan to life. I have modified qglobal.h to ignore -fPIE flag
Sebastian Jennen
@zebastian
Thanks, well done.
Krzysztof Marczak
@buddhi1980
This is only workaround. I waiting for final fix
By the way, adding of saved window layouts doesn't work with Gnome desktop
probably menus which are on the top of the screen doens't allow to have QActionWidgets
Sebastian Jennen
@zebastian
Done, See issue...
What about psyriccios
...Pull request buddhi1980/mandelbulber2#214 ?
Krzysztof Marczak
@buddhi1980
I would not pull it. I don't like binaries from untrusted sources.
Sebastian Jennen
@zebastian
havent seen that, ok, i understand
btw: what do you think about using qInfo()? This can be an output channel like qWarning() but will not display misbehaviour, but rather occasional happening things that may be interesting for the user (like "Cancelled window removing")
These are messages which are not qDebug() since they will stay longterm in the code and the user should not be notified in a hard way (popup), but have it in the logs, if he is interested to know...
Krzysztof Marczak
@buddhi1980
We can use QInfo, because we are already use Qt version > 5.5.
I forgot about this functionality
It's already handled by myMessageOutput
Sebastian Jennen
@zebastian
OK great! they are probably some qDebug() that should be changed to qInfo() but this has lowest priority.
Krzysztof Marczak
@buddhi1980
We are really close to release new version. Everything depends how many bugs Graeme will find :smile: Now I'm trying to fix bug #153
Sebastian Jennen
@zebastian
in audio frames per keyframe are loaded as double: gPar->Get<double>("frames_per_keyframe"), should be int?
Krzysztof Marczak
@buddhi1980
it's a mistake
shoudl be Get<int>()
Sebastian Jennen
@zebastian
thats what i thought, got a fix ready, just need some minutes...
Sebastian Jennen
@zebastian
just pushed :)
also put the slotPlayPositionChanged back in seekto, (ping pong issue is fixed with bool updateSlider
Krzysztof Marczak
@buddhi1980
great
Now I'm working on flight animation and gamepad
when I finish my part of changes, you have to check if gamepad works as it should be or to finish needed changes
my gamepad doesn't work with QGamePad, so I cannot test it