Arch Linux

zester

Member

From: Wilkes Barre Pa

Registered: 2011-08-13

Posts: 156

Website

Re: Future of QT and KDE

Gusar, what do you mean by "Nvidia *already* uses kernel modesetting. Have so since forever."?

Exactly that. They use kernel modesetting since forever. Since way before KMS existed.

Yahhh you have no clue what your talking about.

Gusar

Member

Registered: 2009-08-25

Posts: 3,606

Re: Future of QT and KDE

Gusar do you even know why we need GLX for 3D?

I still fail to see what GLX has to do with kernel modesetting.

They're using their own implementation like with libgl?

Bingo!

I'm surprised zester... You appear to be quite a skilled developer. That someone like you will go mix up a concept and an implementation... Like I said, surprised. Karol understood immediately.

Last edited by Gusar (2011-09-07 01:13:53)

karol

Archivist

Registered: 2009-05-06

Posts: 25,440

Re: Future of QT and KDE

They're using their own implementation like with libgl?

Bingo!

I feel it would help a lot if you said it yourself at the beginning :-) I can't blame zester for getting confused.

Gusar

Member

Registered: 2009-08-25

Posts: 3,606

Re: Future of QT and KDE

I feel it would help a lot if you said it yourself at the beginning :-) I can't blame zester for getting confused.

It should have been quite obvious. Especially to a skilled developer.

zester

Member

From: Wilkes Barre Pa

Registered: 2011-08-13

Posts: 156

Website

Re: Future of QT and KDE

Gusar do you even know why we need GLX for 3D?

I still fail to see what GLX has to do with kernel modesetting.

They're using their own implementation like with libgl?

Bingo!

Actually kernel mode settings or mode settings is when the video driver sits directly ontop of the kernel and that is NOT what the nvidia driver does
it works in conjunction with the GLX X extension hence why you need both X and GLX. With kernel mode settings you the nvidia driver would render directly
on the kernel stack and you wouldn't need X or GLX for 3D. and the Wayland Nvidia issue wouldn't exist.

Last edited by zester (2011-09-07 01:16:34)

zester

Member

From: Wilkes Barre Pa

Registered: 2011-08-13

Posts: 156

Website

Re: Future of QT and KDE

A little demonstration for you http://wayland.freedesktop.org/architecture.html

See where the compositor sits.

The X server receives the rendering requests from the compositor and either copies the compositor back buffer to the front buffer or does a pageflip.

Last edited by zester (2011-09-07 01:21:39)

fsckd

Forum Fellow

Registered: 2009-06-15

Posts: 4,173

Re: Future of QT and KDE

Moderator note: Just a quick reminder guys: https://wiki.archlinux.org/index.php/Fo … ther_Users

---

aur S & M :: forum rules :: Community Ethos
Resources for Women, POC, LGBT*, and allies

Gusar

Member

Registered: 2009-08-25

Posts: 3,606

Re: Future of QT and KDE

Actually kernel mode settings or mode settings is when the video driver sits directly ontop of the kernel and that is NOT what the nvidia driver does

I can fetch you quotes form nvidia developers. It does.

With kernel mode settings you the nvidia driver would render directly on the kernel stack and you wouldn't need X or GLX for 3D.

I don't see how one necessarily follows the other. The kernel module sets the mode. Who then uses that is a different matter. Nvidia only has one user of it: X. Open source drivers have three: X, fbcon and Wayland. Nvidia could easily have Wayland as another user. And I'm sure it will, once Wayland matures.

zester

Member

From: Wilkes Barre Pa

Registered: 2011-08-13

Posts: 156

Website

Re: Future of QT and KDE

http://en.wikipedia.org/wiki/Mode-setting

As NVIDIA did not release all the needed documentation for its graphics chip, the development is under the nouveau project which uses reverse engineering to get it to work. Nouveau has been accepted in version 2.6.33 of the kernel which was released on December 10, 2009. This will allow to use kernel-based mode-setting for NVIDIA cards with this driver.

zester

Member

From: Wilkes Barre Pa

Registered: 2011-08-13

Posts: 156

Website

Re: Future of QT and KDE

Mode-setting can be done in kernel space or in user space. The Nvidia binary driver does it in user space and not kernel space. As I said before KMS would be a GPL violation.

The Nvidia Driver does NOT support KMS

Last edited by zester (2011-09-07 01:31:22)

Gusar

Member

Registered: 2009-08-25

Posts: 3,606

Re: Future of QT and KDE

http://en.wikipedia.org/wiki/Mode-setting

As NVIDIA did not release all the needed documentation for its graphics chip, the development is under the nouveau project which uses reverse engineering to get it to work. Nouveau has been accepted in version 2.6.33 of the kernel which was released on December 10, 2009. This will allow to use kernel-based mode-setting for NVIDIA cards with this driver.

You're still mixing up the concept and an implementation. Nouveau uses KMS, which is one possible implementation of doing kernel modesetting. But this has nothing at all to do with the nvidia driver, which uses it's own implementation. And has so since before KMS even existed.

Nvidia had many things before the open source stack got them. Accelerated indirect rendering would be a good example. Nvidia had it way before AIGLX was implemented in the open stack.

Mode-setting can be done in kernel space or in user space. The Nvidia binary driver does it in user space and not KMS

The Nvidia driver does it in the kernel. Not KMS, their own implementation. I never claimed nvidia uses KMS.

Last edited by Gusar (2011-09-07 01:33:37)

zester

Member

From: Wilkes Barre Pa

Registered: 2011-08-13

Posts: 156

Website

Re: Future of QT and KDE

Tell you what for the sake that I might be mixing this stuff up. I will just agree with you until I get my email back from Linux explaining it in detail to me. I would rather talk about Qt anyways

Last edited by zester (2011-09-07 01:33:17)

zester

Member

From: Wilkes Barre Pa

Registered: 2011-08-13

Posts: 156

Website

Re: Future of QT and KDE

And modesetting solution isn't the problem. It's the GPL violation of a Binary Blob that operates on that level. Hence why Nvidia won't touch it.

Nvidia *already* uses kernel modesetting. Have so since forever. So I have no idea what GPL violation you mean here, and what has changed now that Wayland is around.

No you said kernel mode settings KMS

zester

Member

From: Wilkes Barre Pa

Registered: 2011-08-13

Posts: 156

Website

Re: Future of QT and KDE

I am not a linux kernel engineer. kernel modesetting to me means just that KMS. And what I know of the topic is that
there is issues regarding Qt, Wayland and Nvidia Binary Blob that could use some discussion.

And you dragged me way out of my zone of expertise. But thats ok atleast I have something new to research so be prepared for our next
in counter ill be armed with a weath of knowledge waiting for you to make a mistake lol joking.

Last edited by zester (2011-09-07 01:40:29)

Gusar

Member

Registered: 2009-08-25

Posts: 3,606

Re: Future of QT and KDE

kernel modesetting to me means just that KMS

KMS is an implementation. One possible implementation of doing kernel modesetting. Nvidia has their own. Here, directly from a Nvidia developer: http://www.nvnews.net/vbulletin/showthr … ost1946400

ewaller

Administrator

From: Pasadena, CA

Registered: 2009-07-13

Posts: 20,326

Re: Future of QT and KDE

Mode-setting can be done in kernel space or in user space. We are talking about kernel space, there would be no violation in user space. But thats not the same thing as KMS

In case you were wondering what happened to your post... I think you meant to hit the reply link rather than the report link.  It happens more than you could know

Last edited by ewaller (2011-09-07 01:59:41)

---

Nothing is too wonderful to be true, if it be consistent with the laws of nature -- Michael Faraday
Sometimes it is the people no one can imagine anything of who do the things no one can imagine. -- Alan Turing
---
How to Ask Questions the Smart Way

zester

Member

From: Wilkes Barre Pa

Registered: 2011-08-13

Posts: 156

Website

Re: Future of QT and KDE

http://www.nvnews.net/vbulletin/showthread.php?t=129253

KMS symbols are GPL. NVidia can't legally use them. 

Ill find more info something more official.

zester

Member

From: Wilkes Barre Pa

Registered: 2011-08-13

Posts: 156

Website

Re: Future of QT and KDE

Mode-setting can be done in kernel space or in user space. We are talking about kernel space, there would be no violation in user space. But thats not the same thing as KMS

In case you were wondering what happened to your post... I think you meant to hit the reply link rather than the report link.  It happens more than you could know

Lol thanks Man you guys got that forum mod thing down don't you lol

Last edited by zester (2011-09-07 02:14:29)

zester

Member

From: Wilkes Barre Pa

Registered: 2011-08-13

Posts: 156

Website

Re: Future of QT and KDE

I have E-Mails sent out to both Linus and the Wayland mailing list in regards to the Nvidia Binary Blob legal issue I will be back when I have more ammo to
fight a good fight :0

Thanks that session was a good stress re-leaver. Maybe next time we can fling monkey poop at each other lol

Last edited by zester (2011-09-07 02:20:18)

Teho

Member

Registered: 2010-01-30

Posts: 200

Re: Future of QT and KDE

1. They don't have Nokia, Intel devices to develop for anymore <-- Devices as in hardware and they don't. Nokia isn't building anymore devices for Qt or Meego and all those developers came from
Nokia to develop on symbian with Qt but symbian is dead. And I said nothing about anyone getting fired.
2. What does Digia and there bugfixes have to do with anything I said they outsourced commercial offerings to digia.
3. Supposedly but Intels comment to the rumor was that they were going to continue to support Meego and Open Source and not
that there current plans of developing and shipping Meego devices were still in effect. Thats a huge difference. Meego doesn't need Intels code it needs it's devices!

Are you serious? If we are talking the state of developement of Qt toolkit obviosly the most important thing is that it's getting developed and maintained. So what's the problem with Nokia outsourcing their commercial Qt offering if Digia does the same that Nokia did before? Or did you just want to inform me that Nokia outsourced something and nothing changed? If Intel developes MeeGo it also means that they have at least somekind of commercial plan, it's not like they would spend money on something that has absolutely no finansical gain for them ever or would they?

http://labs.qt.nokia.com/2011/05/12/qt- … -the-list/

In blog post you linked:

Anyway, summary is:
a) QtSvg is not getting removed; so stop tearing your hair out.
b) we’re not going to work on it anymore, it will just stay as-is;
c) if you have problems with its SVG support, first check if your source is compliant to SVG Tiny 1.2; if it’s not, use QtWebKit.

-Thiago Macieira (Senior Product Manager for Qt Developer Frameworks)

kachelaqa

Member

Registered: 2010-09-26

Posts: 216

Re: Future of QT and KDE

http://labs.qt.nokia.com/2011/05/12/qt- … -the-list/

In blog post you linked:

Anyway, summary is:
a) QtSvg is not getting removed; so stop tearing your hair out.
b) we’re not going to work on it anymore, it will just stay as-is;
c) if you have problems with its SVG support, first check if your source is compliant to SVG Tiny 1.2; if it’s not, use QtWebKit.

-Thiago Macieira (Senior Product Manager for Qt Developer Frameworks)

there are some other comments in that thread that are even more telling:

Jason May 14, 2011 at 5:27 am
@Thiago: Does it mean that we don’t have a class named ‘QSvg’ any more in QT5?
And people who need to paint a SVG image have to load the whole Qtwebkit lib?

Thiago Macieira May 14, 2011 at 9:04 am
@Jason: to be honest, we’d like to remove it. But given the reaction, we probably won’t. But we won’t do much more than ensure it compiles.

Is there anyone in this 67-comment thread that wants to do some SVG work? I can’t believe that with so many people who want it to happen, no one wants to make it happen.

the "future" of an open source project consists of the people capable and willing to make it happen. the technical details are largely irrelevant.

Last edited by kachelaqa (2011-09-07 11:50:23)

zester

Member

From: Wilkes Barre Pa

Registered: 2011-08-13

Posts: 156

Website

Re: Future of QT and KDE

Here is the response from one of the Wayland developers.

I am trying to figure out what the issue is with Nvidia supporting Wayland for it's Binary Blob.

From my understanding KMS symbols are GPL. NVidia can't legally use them.
but hasn't the Nvidia Binary Driver always used some form of KMS?

Specifically, KMS is a relatively-new API which nvidia has never used.
They have done modesetting in the kernel, but they don't participate
in Kernel Modesetting.

This might seem confusing. Nomenclature's always tricky.

I can forward the e-mail to anyone who wants it.

I am still wait on other emails from kernel developers in regards to MS and KMS, the legal issue with the Nvidia Binary Blob and the GPL.
sooo stay tuned.

Last edited by zester (2011-09-07 12:45:20)

Anonymo

Member

Registered: 2005-04-07

Posts: 427

Website

Re: Future of QT and KDE

I'm not a programmer.  What is the advantage to switching to javascript instead of C++?

zester

Member

From: Wilkes Barre Pa

Registered: 2011-08-13

Posts: 156

Website

Re: Future of QT and KDE

I'm not a programmer.  What is the advantage to switching to javascript instead of C++?

Are you referring to the v8 javascript engine I was talking about earlyer?

Actually I use C++ with Javascript.

Lets say I write a FileManager and it is designed to allow non c++ programmers to
write extension(plugins) for it. Well the FilManager it self would be written in C++ and I would
embed a Javascript engine inside and then expose some javascript functions to certain C++
functions. You could then load this javascript file that the internal javascript engine
would interpret calling the c++ functions that correspond to the javascript functions that your using.

I use C++ and Javascript because those to languages
alone cover every aspect of programming I am interested in.

Application, Game, Multimedia, Web Development.

I can use just javascript for Server and Client Side
scripting in terms of Web Development.

And the Qt toolkit makes this easy for use to do.

Last edited by zester (2011-09-07 15:21:19)

zester

Member

From: Wilkes Barre Pa

Registered: 2011-08-13

Posts: 156

Website

Re: Future of QT and KDE

Ok I have my coffee, good night sleep I am ready to start the pissing contest .

I am not saying that any of this is a guarantee, just that the future for Ati/Nvidia binary blobs isn't looking good in regards to Qt and highpoly game development.

QtSvg
Overall module state: Deprecated
New maintainer required
Reasoning: SVG Full (as opposed to SVG Tiny) functionality available in QtWebKit, which should be used instead; we welcome research for a replacement for the SVG-generating code.

Note: SVG Full (as opposed to SVG Tiny) functionality available in QtWebKit, which should be used instead;

QtSvg is one of the modules that will not see any further development, And is being offloaded to the community. Depending on how bad binary compatibility is broken in Qt5 and
if the QtSvg module isnt updated the QtSvg module won't work with Qt5.

@Jason: to be honest, we’d like to remove it. But given the reaction, we probably won’t. But we won’t do much more than ensure it compiles.

Anyway, summary is:
a) QtSvg is not getting removed; so stop tearing your hair out.
b) we’re not going to work on it anymore, it will just stay as-is;
c) if you have problems with its SVG support, first check if your source is compliant to SVG Tiny 1.2; if it’s not, use QtWebKit.

As I said before in regards to QtSvg, Guess what QtWebKit is no replacement when developing a Drawing application that saves as Svg or Rendering Svg graphics on the desktop.

Steven June 6, 2011 at 6:39 am

I just don’t get all this Wayland hype, for it to work properly on Nvidia cards, Nvidia would have to add some code to the kernel to support kernel mode setting, which would require the Nvidia driver to be GPL. This will never ever happen. Wayland is vapor-ware just forget about it.

NVidia is not the only video card manufacturer. And their proprietary driver not even the only option for their own cards.

Besides, NVidia cards already support embedded systems without X, albeit with other solutions besides Wayland. And they do this without KMS.

Finally, vapourware is software that never becomes concrete. You cannot tell the people who are actually doing the work that their work is vapour. That’s just plain insulting. I’ve seen it running, I’ve worked with the people making a reality. What is your contribution?

I’ve been contributing to Linux and KDE opensource software development for as long as KDE has existed.

My point is the opensource community finally has AAA game development library’s, that need those proprietary Nvidia drivers to compete on the same level as other platforms.

I would absolutely love to see a modern replacement for X even if that meant we had to start from scratch and rewrite all legacy applications, not saying we would.

But I have personally spoken with the Nvidia devs that produce the linux driver, and there just not going to go for wayland. And the Intel, Nouveau driver’s just don’t come close to the
proprietary AMD/Ati and Nvidia offerings.

“NVidia cards already support embedded systems without X”
You mean on Mesa without hardware acceleration? Yah Ive seen that trick also.

No, I meant full hardware-accelerated, non-X NVidia systems (Tegra II). It exists. It might not be freely available though.

Anyway, the point is that we’re not stopping for “I don’t have a driver”. We are going to optimise for the case where you do.

And modesetting solution isn't the problem. It's the GPL violation of a Binary Blob that operates on that level. Hence why Nvidia won't touch it.

Nvidia *already* uses kernel modesetting. Have so since forever. So I have no idea what GPL violation you mean here, and what has changed now that Wayland is around.

No you said kernel mode settings KMS

I am trying to figure out what the issue is with Nvidia supporting Wayland for it's Binary Blob.

From my understanding KMS symbols are GPL. NVidia can't legally use them.
but hasn't the Nvidia Binary Driver always used some form of KMS?

Specifically, KMS is a relatively-new API which nvidia has never used.
They have done modesetting in the kernel, but they don't participate
in Kernel Modesetting.

This might seem confusing. Nomenclature's always tricky.

I proved my point on the QtSvg issue and and 1/2 half of my point in regards to Waland and the Nividia Driver and Light House. Once I get those emails back confirming
my suspecion on the Nvidia Binary Blob KMS GPL violation I can move on to the X vs Wayland/Qt5 ordeal.

Last edited by zester (2011-09-07 15:54:01)