Re: [Development] QtMultimedia && QML - hard performance regressions on Windows

2016-04-29 Thread Denis Shienkov
Seems, the bug is in this two patches:

   - 36549dbe148055d6ecf98952b05b4ff2310fe491 - DirectShow: use the EVR in
   the renderer control.
   - c7397523e77578cf8f09ba3258791f68c34e2e5f - Windows: Improve EVR
   presenter.

I have added this info to bug-tracker...



2016-04-29 11:51 GMT+03:00 Denis Shienkov :

> I have checked it again on other PC and see following:
>
> 1) If I compile the QtMultiMedia from the 5.6 branch (self-compiled with
> the Qt 5.6.0) and there are the same issue!!!
>
> 2) But if I use QtMultiMedia from the 5.5 branch (self-compiled with the
> Qt 5.6.0), then this issue does not exists!!!
>
> PS: I have added this info to the bug-tracker too
>
>
>
> 2016-04-28 22:25 GMT+03:00 Marc Mutz :
>
>> On Thursday 28 April 2016 20:24:45 Denis Shienkov wrote:
>> > I'm currently have checked it with the Qt 5.5 and have no this
>> > problem... So, the regression was introduced in Qt 5.6...
>> >
>> > So, Qt-developers, do you checks your code of QtMultimedia on
>> > regressions when you accept a patches, and before releasing? I do not
>> > understand, how is it possible to introduce this epic regression??? How
>> > is it possible to work for a salary and to make such bugs (and, please
>> > do not say that it is open source)?
>>
>> git bisect start origin/5.6 origin/5.5
>>
>> --
>> Marc Mutz  | Senior Software Engineer
>> KDAB (Deutschland) GmbH & Co.KG, a KDAB Group Company
>> Tel: +49-30-521325470
>> KDAB - Qt, C++ and OpenGL Experts
>>
>
>
___
Development mailing list
Development@qt-project.org
http://lists.qt-project.org/mailman/listinfo/development


Re: [Development] QtMultimedia && QML - hard performance regressions on Windows

2016-04-29 Thread Denis Shienkov
I have checked it again on other PC and see following:

1) If I compile the QtMultiMedia from the 5.6 branch (self-compiled with
the Qt 5.6.0) and there are the same issue!!!

2) But if I use QtMultiMedia from the 5.5 branch (self-compiled with the Qt
5.6.0), then this issue does not exists!!!

PS: I have added this info to the bug-tracker too



2016-04-28 22:25 GMT+03:00 Marc Mutz :

> On Thursday 28 April 2016 20:24:45 Denis Shienkov wrote:
> > I'm currently have checked it with the Qt 5.5 and have no this
> > problem... So, the regression was introduced in Qt 5.6...
> >
> > So, Qt-developers, do you checks your code of QtMultimedia on
> > regressions when you accept a patches, and before releasing? I do not
> > understand, how is it possible to introduce this epic regression??? How
> > is it possible to work for a salary and to make such bugs (and, please
> > do not say that it is open source)?
>
> git bisect start origin/5.6 origin/5.5
>
> --
> Marc Mutz  | Senior Software Engineer
> KDAB (Deutschland) GmbH & Co.KG, a KDAB Group Company
> Tel: +49-30-521325470
> KDAB - Qt, C++ and OpenGL Experts
>
___
Development mailing list
Development@qt-project.org
http://lists.qt-project.org/mailman/listinfo/development


Re: [Development] QtMultimedia && QML - hard performance regressions on Windows

2016-04-28 Thread Marc Mutz
On Thursday 28 April 2016 20:24:45 Denis Shienkov wrote:
> I'm currently have checked it with the Qt 5.5 and have no this 
> problem... So, the regression was introduced in Qt 5.6...
> 
> So, Qt-developers, do you checks your code of QtMultimedia on 
> regressions when you accept a patches, and before releasing? I do not 
> understand, how is it possible to introduce this epic regression??? How 
> is it possible to work for a salary and to make such bugs (and, please 
> do not say that it is open source)?

git bisect start origin/5.6 origin/5.5

-- 
Marc Mutz  | Senior Software Engineer
KDAB (Deutschland) GmbH & Co.KG, a KDAB Group Company
Tel: +49-30-521325470
KDAB - Qt, C++ and OpenGL Experts
___
Development mailing list
Development@qt-project.org
http://lists.qt-project.org/mailman/listinfo/development


Re: [Development] QtMultimedia && QML - hard performance regressions on Windows

2016-04-28 Thread Denis Shienkov
I'm currently have checked it with the Qt 5.5 and have no this 
problem... So, the regression was introduced in Qt 5.6...


So, Qt-developers, do you checks your code of QtMultimedia on 
regressions when you accept a patches, and before releasing? I do not 
understand, how is it possible to introduce this epic regression??? How 
is it possible to work for a salary and to make such bugs (and, please 
do not say that it is open source)?



27.04.2016 16:44, Massimo Callegari via Development пишет:


Da: Mitch Curtis 
A: Massimo Callegari ; "development@qt-project.org" 

Inviato: Mercoledì 27 Aprile 2016 14:56
Oggetto: RE: [Development] QtMultimedia && QML - hard performance regressions 
onWindows




I didn't look at the other bugs, but QTBUG-37004 seems like it would pretty 
easy to contribute a fix yourself.

I don't believe it is pretty easy, especially on Linux.
OSX already returns a user friendly list of devices.
Windows cuts the devices names to 31 characters, cause it use APIs coming 
directly from the 90s.
Linux is the worst case instead, cause every possible ALSA device is returned, 
so if you have 2 sound cards it will return like 20 devices, most of which 
don't even work.
I believe users expect to see a list like what KDE/Gnome expose, so 4-5 devices 
total.
Plus I should be looking at the pulseaudio implementation as well, which I know 
nothing about.
In any case, it would require an internal change of the logic, most likely 
breaking backward compatibility, which is not something that I can decide.


Why waste your time and energy being outraged when you could be fixing it? 
Seems like a far more effective way of getting things fixed in an open source 
project.

Well, I believe reporting an issue is already a contribution to fixing it and while a 
"pretty easy" issue would take me 1 hour to be fixed, the overall effort to get 
it approved would be 100 hours. I already went through the process for a QtSerialPort 
method, and it was a hell of a journey. Especially working with Gerrit. Please excuse my 
limited GIT knowledge :(


As said in the bug report, higher priority issues are being worked on.

Which one ?
And also...since 1-2 years ? Right...


I would rather appreciate much more if the Qt Company could start a sort of "bounty 
program", where users can place a few bucks for the resolution of specific issues.
There are more and more examples of this mechanism, like Kickstarter, 
Bountysource or even Stackoverflow.
I would be happy to participate to it for the most urgent issues affecting my 
projects.
I'm pretty sure there are a lot of open source projects raising donations, and 
small companies willing to spend some money on issues that are a priority for 
them.

Instead, it seems to me (and it's purely my mere assumption) that big bucks 
come from big companies and they have the full attention of the Qt developments.
Attention that don't go to all the rest of minor developers/projects/companies 
that are actually the largest user base of the Qt libraries.
If this is really the truth, it is pretty sad. Again, just my opinion.



Da: Denis Shienkov 
A: "development@qt-project.org" 
Inviato: Mercoledì 27 Aprile 2016 11:37
Oggetto: [Development] QtMultimedia && QML - hard performance regressions
onWindows



Hi developers.

I faced with a big problem with the video playback performance, when I
try to open the mp4 video file from any of QML video examples. I have a
high CPU loading (~40%) and the video is freezing and jerks. But the C++
media player works satisfactory, more or less tolerably (on point 3 from
the 1-5 points)...

It is impossible to use at all in QML, I face more and more with the
surprises of QtMultimedia, and disappointed more and more...

WTF ? What sense in QtMultimedia if it does not work as expected?

PS: I have created a bug: https://bugreports.qt.io/browse/QTBUG-53019

BR,
Denis


___
Development mailing list
Development@qt-project.org
http://lists.qt-project.org/mailman/listinfo/development
___
Development mailing list
Development@qt-project.org
http://lists.qt-project.org/mailman/listinfo/development

___
Development mailing list
Development@qt-project.org
http://lists.qt-project.org/mailman/listinfo/development


___
Development mailing list
Development@qt-project.org
http://lists.qt-project.org/mailman/listinfo/development


Re: [Development] QtMultimedia && QML - hard performance regressions on Windows

2016-04-27 Thread Massimo Callegari via Development


Da: Mitch Curtis 
A: Massimo Callegari ; "development@qt-project.org" 
 
Inviato: Mercoledì 27 Aprile 2016 14:56
Oggetto: RE: [Development] QtMultimedia && QML - hard performance regressions 
onWindows



>I didn't look at the other bugs, but QTBUG-37004 seems like it would pretty 
>easy to contribute a fix yourself. 

I don't believe it is pretty easy, especially on Linux. 
OSX already returns a user friendly list of devices.
Windows cuts the devices names to 31 characters, cause it use APIs coming 
directly from the 90s.
Linux is the worst case instead, cause every possible ALSA device is returned, 
so if you have 2 sound cards it will return like 20 devices, most of which 
don't even work.
I believe users expect to see a list like what KDE/Gnome expose, so 4-5 devices 
total.
Plus I should be looking at the pulseaudio implementation as well, which I know 
nothing about.
In any case, it would require an internal change of the logic, most likely 
breaking backward compatibility, which is not something that I can decide.

> Why waste your time and energy being outraged when you could be fixing it? 
> Seems like a far more effective way of getting things fixed in an open source 
> project. 

Well, I believe reporting an issue is already a contribution to fixing it and 
while a "pretty easy" issue would take me 1 hour to be fixed, the overall 
effort to get it approved would be 100 hours. I already went through the 
process for a QtSerialPort method, and it was a hell of a journey. Especially 
working with Gerrit. Please excuse my limited GIT knowledge :(

> As said in the bug report, higher priority issues are being worked on.

Which one ?
And also...since 1-2 years ? Right...


I would rather appreciate much more if the Qt Company could start a sort of 
"bounty program", where users can place a few bucks for the resolution of 
specific issues.
There are more and more examples of this mechanism, like Kickstarter, 
Bountysource or even Stackoverflow.
I would be happy to participate to it for the most urgent issues affecting my 
projects.
I'm pretty sure there are a lot of open source projects raising donations, and 
small companies willing to spend some money on issues that are a priority for 
them.

Instead, it seems to me (and it's purely my mere assumption) that big bucks 
come from big companies and they have the full attention of the Qt developments.
Attention that don't go to all the rest of minor developers/projects/companies 
that are actually the largest user base of the Qt libraries.
If this is really the truth, it is pretty sad. Again, just my opinion.

> 
> 
> Da: Denis Shienkov 
> A: "development@qt-project.org" 
> Inviato: Mercoledì 27 Aprile 2016 11:37
> Oggetto: [Development] QtMultimedia && QML - hard performance regressions
> onWindows
> 
> 
> 
> Hi developers.
> 
> I faced with a big problem with the video playback performance, when I
> try to open the mp4 video file from any of QML video examples. I have a
> high CPU loading (~40%) and the video is freezing and jerks. But the C++
> media player works satisfactory, more or less tolerably (on point 3 from
> the 1-5 points)...
> 
> It is impossible to use at all in QML, I face more and more with the
> surprises of QtMultimedia, and disappointed more and more...
> 
> WTF ? What sense in QtMultimedia if it does not work as expected?
> 
> PS: I have created a bug: https://bugreports.qt.io/browse/QTBUG-53019
> 
> BR,
> Denis
> 
> 
> ___
> Development mailing list
> Development@qt-project.org
> http://lists.qt-project.org/mailman/listinfo/development
> ___
> Development mailing list
> Development@qt-project.org
> http://lists.qt-project.org/mailman/listinfo/development
___
Development mailing list
Development@qt-project.org
http://lists.qt-project.org/mailman/listinfo/development


Re: [Development] QtMultimedia && QML - hard performance regressions on Windows

2016-04-27 Thread Mitch Curtis


> -Original Message-
> From: Development [mailto:development-bounces+mitch.curtis=qt.io@qt-
> project.org] On Behalf Of Massimo Callegari via Development
> Sent: Wednesday, 27 April 2016 12:28 PM
> To: Denis Shienkov ; development@qt-project.org
> Subject: Re: [Development] QtMultimedia && QML - hard performance
> regressions on Windows
> 
> Hey Denis, welcome to the "QtMultimedia lovers" club :)
> 
> Good luck with your remarks, most likely nobody will care.
> It seems that the primary role of QtMultimedia is to capture input video.
> All the rest doesn't matter.
> 
> I've opened a number of tickets and none of them have been solved:
> https://bugreports.qt.io/browse/QTBUG-37004  (Feb 2014)
> https://bugreports.qt.io/browse/QTBUG-37005  (Feb 2014)
> https://bugreports.qt.io/browse/QTBUG-37882  (Mar 2014)
> https://bugreports.qt.io/browse/QTBUG-42034  (Oct 2014)
> https://bugreports.qt.io/browse/QTBUG-47284  (Jul 2015)
> 
> 3 on 5 marked as P2.
> 
> I would also add a recent discovery of how non-cross platform QtMultimedia
> is.
> The use case is starting a video in fullscreen on a second monitor
> (QtWidget way).
> In Linux and OSX you have to do
> m_videoWidget->setGeometry(rect);
> m_videoWidget->setFullScreen(true);and on Windows m_videoWidget-
> >setFullScreen(true);
> m_videoWidget->setGeometry(rect);
> 
> I didn't even open a ticket for that, cause I got tired to waste my time
> on this topic.

I didn't look at the other bugs, but QTBUG-37004 seems like it would pretty 
easy to contribute a fix yourself. Why waste your time and energy being 
outraged when you could be fixing it? Seems like a far more effective way of 
getting things fixed in an open source project. As said in the bug report, 
higher priority issues are being worked on.

> 
> 
> Da: Denis Shienkov 
> A: "development@qt-project.org" 
> Inviato: Mercoledì 27 Aprile 2016 11:37
> Oggetto: [Development] QtMultimedia && QML - hard performance regressions
> onWindows
> 
> 
> 
> Hi developers.
> 
> I faced with a big problem with the video playback performance, when I
> try to open the mp4 video file from any of QML video examples. I have a
> high CPU loading (~40%) and the video is freezing and jerks. But the C++
> media player works satisfactory, more or less tolerably (on point 3 from
> the 1-5 points)...
> 
> It is impossible to use at all in QML, I face more and more with the
> surprises of QtMultimedia, and disappointed more and more...
> 
> WTF ? What sense in QtMultimedia if it does not work as expected?
> 
> PS: I have created a bug: https://bugreports.qt.io/browse/QTBUG-53019
> 
> BR,
> Denis
> 
> 
> ___
> Development mailing list
> Development@qt-project.org
> http://lists.qt-project.org/mailman/listinfo/development
> ___
> Development mailing list
> Development@qt-project.org
> http://lists.qt-project.org/mailman/listinfo/development
___
Development mailing list
Development@qt-project.org
http://lists.qt-project.org/mailman/listinfo/development


Re: [Development] QtMultimedia && QML - hard performance regressions on Windows

2016-04-27 Thread Denis Shienkov
> I've seen something similar and solved it by changing codecs (installed
k-lite
codec pack)

I have the K-Lite codec pack already installed (I'm sorry, if I haven't
mentioned it).

BR,
Denis

2016-04-27 14:10 GMT+03:00 Sergio Martins :

> On Wednesday, 27 April 2016 12:37:26 WEST Denis Shienkov wrote:
> > Hi developers.
> >
> > I faced with a big problem with the video playback performance, when I
> try
> > to open the mp4 video file from any of QML video examples. I have a high
> > CPU loading (~40%) and the video is freezing and jerks. But the C++ media
> > player works satisfactory, more or less tolerably (on point 3 from the
> 1-5
> > points)...
>
> I've seen something similar and solved it by changing codecs (installed
> k-lite
> codec pack)
>
>
> Regards,
> --
> Sérgio Martins | sergio.mart...@kdab.com | Software Engineer
> Klarälvdalens Datakonsult AB, a KDAB Group company
> Tel: Sweden (HQ) +46-563-540090, USA +1-866-777-KDAB(5322)
> KDAB - The Qt Experts
>
___
Development mailing list
Development@qt-project.org
http://lists.qt-project.org/mailman/listinfo/development


Re: [Development] QtMultimedia && QML - hard performance regressions on Windows

2016-04-27 Thread Sergio Martins
On Wednesday, 27 April 2016 12:37:26 WEST Denis Shienkov wrote:
> Hi developers.
> 
> I faced with a big problem with the video playback performance, when I try
> to open the mp4 video file from any of QML video examples. I have a high
> CPU loading (~40%) and the video is freezing and jerks. But the C++ media
> player works satisfactory, more or less tolerably (on point 3 from the 1-5
> points)...

I've seen something similar and solved it by changing codecs (installed k-lite 
codec pack)


Regards,
-- 
Sérgio Martins | sergio.mart...@kdab.com | Software Engineer
Klarälvdalens Datakonsult AB, a KDAB Group company
Tel: Sweden (HQ) +46-563-540090, USA +1-866-777-KDAB(5322)
KDAB - The Qt Experts
___
Development mailing list
Development@qt-project.org
http://lists.qt-project.org/mailman/listinfo/development


Re: [Development] QtMultimedia && QML - hard performance regressions on Windows

2016-04-27 Thread Massimo Callegari via Development
Hey Denis, welcome to the "QtMultimedia lovers" club :)

Good luck with your remarks, most likely nobody will care. 
It seems that the primary role of QtMultimedia is to capture input video. All 
the rest doesn't matter.

I've opened a number of tickets and none of them have been solved:
https://bugreports.qt.io/browse/QTBUG-37004  (Feb 2014)
https://bugreports.qt.io/browse/QTBUG-37005  (Feb 2014)
https://bugreports.qt.io/browse/QTBUG-37882  (Mar 2014)
https://bugreports.qt.io/browse/QTBUG-42034  (Oct 2014)
https://bugreports.qt.io/browse/QTBUG-47284  (Jul 2015)

3 on 5 marked as P2.

I would also add a recent discovery of how non-cross platform QtMultimedia is.
The use case is starting a video in fullscreen on a second monitor (QtWidget 
way).
In Linux and OSX you have to do
m_videoWidget->setGeometry(rect);
m_videoWidget->setFullScreen(true);and on Windows
m_videoWidget->setFullScreen(true);
m_videoWidget->setGeometry(rect);

I didn't even open a ticket for that, cause I got tired to waste my time on 
this topic.



Da: Denis Shienkov 
A: "development@qt-project.org"  
Inviato: Mercoledì 27 Aprile 2016 11:37
Oggetto: [Development] QtMultimedia && QML - hard performance regressions on
Windows



Hi developers.

I faced with a big problem with the video playback performance, when I 
try to open the mp4 video file from any of QML video examples. I have a high 
CPU loading (~40%) and the video is freezing and jerks. But the C++ media 
player works satisfactory, more or less tolerably (on point 3 from the 1-5 
points)...

It is impossible to use at all in QML, I face more and more with the surprises 
of QtMultimedia, and disappointed more and more...  

WTF ? What sense in QtMultimedia if it does not work as expected?

PS: I have created a bug: https://bugreports.qt.io/browse/QTBUG-53019

BR,
Denis


___
Development mailing list
Development@qt-project.org
http://lists.qt-project.org/mailman/listinfo/development
___
Development mailing list
Development@qt-project.org
http://lists.qt-project.org/mailman/listinfo/development