Hi (from the evil KDE SIG;-), we're going to update KDE Plasma Workspaces in Fedora 13 to 4.5.x series. It took some time but we're now pretty sure that it's in a good condition.
The question is (we agreed on KDE SIG meeting yesterday) - should we update Qt to 4.7 too or build KDE stack with current 4.6 series? As there are a few Qt packages outside of KDE SIG/Qt maintainers scope, we'd like to hear any objections against update - bugs we can fix etc. Qt 4.7 is quite well tested, thanks to work on Fedora 14 (Qt 4.7 is already included) and a lot of users are actually using this combination in Fedora 13.
If I put on my Fedora WebKit SIG hat - there are two big +++ for update: 1. newer QtWebKit makes browser's user experiences much more better 2. it's going to be much more easier to backport security issues to Fedora 13 and we care of security issues - WebKit can be quite vulnerable...
Most changes in Qt 4.7 are in mobile development area and new features like Qt Quick.
On the other hand, it could slow down KDE update and it will need more rebuilds.
Is it worth? Any objections from Qt devels? Please comment...
Jaroslav
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
On Wed, 20 Oct 2010 14:02:49 +0200, you wrote:
The question is (we agreed on KDE SIG meeting yesterday) - should we update Qt to 4.7 too or build KDE stack with current 4.6 series? As there are a few Qt packages outside of KDE SIG/Qt maintainers scope, we'd like to hear any objections against update - bugs we can fix etc. Qt 4.7 is quite well tested, thanks to work on Fedora 14 (Qt 4.7 is already included) and a lot of users are actually using this combination in Fedora 13.
Is there a version of PyQt built agains qt-4.7. If so, I can build my application which are depending on it agains the new release. I think there should no issue to build my applications agains a new qt release.
Best Regards:
Jochen Schmitt
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
On Wed, 20 Oct 2010 16:41:16 +0200, you wrote:
Is there a version of PyQt built agains qt-4.7. If so, I can build my application which are depending on it agains the new release. I think there should no issue to build my applications agains a new qt release.
Sorry, the appication which I'M maintaining use the obsolte qt3, so this sentence make no sense.
Due a google search I have found a bug agains stellarium which should be fixed after upgrading to qt-4.7.
Best Regards:
Jochen Schmitt
On 10/20/2010 03:02 PM, Jaroslav Reznik wrote:
The question is (we agreed on KDE SIG meeting yesterday) - should we update Qt to 4.7 too or build KDE stack with current 4.6 series? As there are a few Qt packages outside of KDE SIG/Qt maintainers scope, we'd like to hear any objections against update - bugs we can fix etc. Qt 4.7 is quite well tested, thanks to work on Fedora 14 (Qt 4.7 is already included) and a lot of users are actually using this combination in Fedora 13.
KDE is pretty much self contained, whereas a Qt upgrade affects a much larger number of packages. I don't think updating Qt to a new major version in a stable Fedora release is a good idea; it just causes too much churn.
I am sure there will be bugs in other packages which need to be fixed to work with Qt 4.7. Some of these bugs are going to be fixed in next upstream releases, so people would just need to upgrade their packages and build for F13. Other packages are going to work fine, but fail to rebuild against Qt 4.7 if someone tries (consider security updates: a build which starts suddenly failing when you need to get a two-line security fix quickly out is not so fun).
I personally would not call it a stable release any more: a release where lots of the apps need to be rebuilt, and lots of other apps need to go through integration testing again is not a stable release. It's a rolling development release.
In my opinion a Qt 4.7 update is pure F-14 material. As you say, it is working pretty well on Fedora 14 which is going to be released very soon, so why is there need for Qt 4.7 on F-13?
Thanks for all the nice work on Fedora 14 by the way!
On Sun, 24 Oct 2010 21:24:30 +0300 Kalev Lember kalev@smartlink.ee wrote:
On 10/20/2010 03:02 PM, Jaroslav Reznik wrote:
The question is (we agreed on KDE SIG meeting yesterday) - should we update Qt to 4.7 too or build KDE stack with current 4.6 series? As there are a few Qt packages outside of KDE SIG/Qt maintainers scope, we'd like to hear any objections against update - bugs we can fix etc. Qt 4.7 is quite well tested, thanks to work on Fedora 14 (Qt 4.7 is already included) and a lot of users are actually using this combination in Fedora 13.
KDE is pretty much self contained, whereas a Qt upgrade affects a much larger number of packages. I don't think updating Qt to a new major version in a stable Fedora release is a good idea; it just causes too much churn.
...snip...
I agree with Kalev here. Qt upgrade in a stable release is to be avoided unless there's some severe bug or security issue that can't be backported.
kevin
2010/10/25 Kevin Fenzi kevin@scrye.com
On Sun, 24 Oct 2010 21:24:30 +0300 Kalev Lember kalev@smartlink.ee wrote:
On 10/20/2010 03:02 PM, Jaroslav Reznik wrote:
The question is (we agreed on KDE SIG meeting yesterday) - should we update Qt to 4.7 too or build KDE stack with current 4.6 series? As there are a few Qt packages outside of KDE SIG/Qt maintainers scope, we'd like to hear any objections against update - bugs we can fix etc. Qt 4.7 is quite well tested, thanks to work on Fedora 14 (Qt 4.7 is already included) and a lot of users are actually using this combination in Fedora 13.
KDE is pretty much self contained, whereas a Qt upgrade affects a much larger number of packages. I don't think updating Qt to a new major version in a stable Fedora release is a good idea; it just causes too much churn.
...snip...
I agree with Kalev here. Qt upgrade in a stable release is to be avoided unless there's some severe bug or security issue that can't be backported.
kevin
-- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
I have KDE 4.5.2 in my Fedora 13 wich as far as I understand uses the last version of Qt, the computer is working Flawlessly :D
On Tuesday, October 26, 2010 06:58:52 am Manuel Escudero wrote:
2010/10/25 Kevin Fenzi kevin@scrye.com
On Sun, 24 Oct 2010 21:24:30 +0300
Kalev Lember kalev@smartlink.ee wrote:
On 10/20/2010 03:02 PM, Jaroslav Reznik wrote:
The question is (we agreed on KDE SIG meeting yesterday) - should we update Qt to 4.7 too or build KDE stack with current 4.6 series? As there are a few Qt packages outside of KDE SIG/Qt maintainers scope, we'd like to hear any objections against update - bugs we can fix etc. Qt 4.7 is quite well tested, thanks to work on Fedora 14 (Qt 4.7 is already included) and a lot of users are actually using this combination in Fedora 13.
KDE is pretty much self contained, whereas a Qt upgrade affects a much larger number of packages. I don't think updating Qt to a new major version in a stable Fedora release is a good idea; it just causes too much churn.
...snip...
I agree with Kalev here. Qt upgrade in a stable release is to be avoided unless there's some severe bug or security issue that can't be backported.
kevin
-- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
I have KDE 4.5.2 in my Fedora 13 wich as far as I understand uses the last version of Qt, the computer is working Flawlessly :D
Problem is not KDE itself but there are some more Qt apps in Fedora. And we want to be sure these apps are not affected by update. We are practically 100% sure (as we have a quite good process behind) - but you never know, so we asked for help. And we can help in case of issues. Final Qt is know for 100% ABI (not snapshots :D) compatibility, it was commercial product one year ago...
Jaroslav
On Tuesday, October 26, 2010 06:33:09 am Kevin Fenzi wrote:
On Sun, 24 Oct 2010 21:24:30 +0300
Kalev Lember kalev@smartlink.ee wrote:
On 10/20/2010 03:02 PM, Jaroslav Reznik wrote:
The question is (we agreed on KDE SIG meeting yesterday) - should we update Qt to 4.7 too or build KDE stack with current 4.6 series? As there are a few Qt packages outside of KDE SIG/Qt maintainers scope, we'd like to hear any objections against update - bugs we can fix etc. Qt 4.7 is quite well tested, thanks to work on Fedora 14 (Qt 4.7 is already included) and a lot of users are actually using this combination in Fedora 13.
KDE is pretty much self contained, whereas a Qt upgrade affects a much larger number of packages. I don't think updating Qt to a new major version in a stable Fedora release is a good idea; it just causes too much churn.
...snip...
I agree with Kalev here. Qt upgrade in a stable release is to be avoided unless there's some severe bug or security issue that can't be backported.
Top problem is QtWebKit - it's practically impossible to backport security issues and as everybody knows - WebKit is security issue itself... It's really time consuming task. We hoped we can get standalone QtWebKit by F14/Qt 4.7 but upstream is unable to allocate enough resources to support it now.
We're even considering possibility to split QtWebKit from main Qt package and build it over the older Qt releases but don't like this idea - it probably will bring much more issues than fix (older Qt releases should be supported but no-one tests it probably, so problems can't be avoided here).
For Qt itself - once QtWebKit will be separated, need for update would go to zero... As the most work is done now on mobile devices support. Only one interesting feature for Fedora users has been completed - Qt Quick - it opens a lot of new possibilities for rich desktop application development and I think it should be path for Fedora too (ala for installer, firstboot, some welcome screen etc.).
So with WebKit SIG hat - newer QtWebKit is the must now.
Jaroslav
kevin
On Sunday 24 October 2010 20:24:30 Kalev Lember wrote:
KDE is pretty much self contained, whereas a Qt upgrade affects a much larger number of packages. I don't think updating Qt to a new major version in a stable Fedora release is a good idea; it just causes too much churn.
Nokia managed to upgrade Qt to 4.7 in their Maemo distribution and it got pushed to all devices without causing any problems so far. Their standards for avoiding churn are pretty high and their update scheme is extremely conservative for stable releases. Nevertheless they updated Qt. But they have a pretty good reason for doing that (aligning with future versions of MeeGo and Symbian). So what does a F13 user gain from an upgrade? Is it worth the risks?
F13 isn't what bleeding-edge users are likely to run in the future. Those can easily upgrade to F14 and enjoy the latest stuff. So it's not like they are forced to run a periodical broken rawhide with no security support if they want recent software. I like the idea of Fn getting major updates whereas Fn-1 (that's what F13 is very soon) only gets those updates which are needed for fixing bugs and security issues.
So if the open issues regarding QtWebkit can be solved I agree that leaving Qt at 4.6.x is just fine. If not there ain't much choice as it is pretty much guaranteed that Webkit will have security issues which are mandatory to fix. If upstream only supports 4.7 and backporting isn't an option (which seems to be the case according to jreznik) Qt 4.6 has to go unless some other solution can be implemented before F13 goes EOL. ;)
Lars
On Tue, Oct 26, 2010 at 6:01 PM, Lars Seipel lars.seipel@googlemail.com wrote:
On Sunday 24 October 2010 20:24:30 Kalev Lember wrote:
KDE is pretty much self contained, whereas a Qt upgrade affects a much larger number of packages. I don't think updating Qt to a new major version in a stable Fedora release is a good idea; it just causes too much churn.
Nokia managed to upgrade Qt to 4.7 in their Maemo distribution and it got pushed to all devices without causing any problems so far. Their standards for avoiding churn are pretty high and their update scheme is extremely conservative for stable releases. Nevertheless they updated Qt. But they have a pretty good reason for doing that (aligning with future versions of MeeGo and Symbian). So what does a F13 user gain from an upgrade? Is it worth the risks?
QT isn't the default toolkit in Maemo and it was only introduced at all in the PR1.2 release which only came out around 3-4 months ago so its not a core part of their UI experience on maemo.
So that's not really a good argument for upgrading it in F-13.
Peter
Peter Robinson (pbrobinson@gmail.com) said:
Nokia managed to upgrade Qt to 4.7 in their Maemo distribution and it got pushed to all devices without causing any problems so far. Their standards for avoiding churn are pretty high and their update scheme is extremely conservative for stable releases. Nevertheless they updated Qt. But they have a pretty good reason for doing that (aligning with future versions of MeeGo and Symbian). So what does a F13 user gain from an upgrade? Is it worth the risks?
QT isn't the default toolkit in Maemo and it was only introduced at all in the PR1.2 release which only came out around 3-4 months ago so its not a core part of their UI experience on maemo.
So that's not really a good argument for upgrading it in F-13.
Also, I'd assume that Maemo explictily ships/supports a much smaller set of software than Fedora, so they can explicitly list all their dependencies that may need fixed.
Bill
On Tuesday, October 26, 2010 08:39:07 pm Bill Nottingham wrote:
Peter Robinson (pbrobinson@gmail.com) said:
Nokia managed to upgrade Qt to 4.7 in their Maemo distribution and it got pushed to all devices without causing any problems so far. Their standards for avoiding churn are pretty high and their update scheme is extremely conservative for stable releases. Nevertheless they updated Qt. But they have a pretty good reason for doing that (aligning with future versions of MeeGo and Symbian). So what does a F13 user gain from an upgrade? Is it worth the risks?
QT isn't the default toolkit in Maemo and it was only introduced at all in the PR1.2 release which only came out around 3-4 months ago so its not a core part of their UI experience on maemo.
So that's not really a good argument for upgrading it in F-13.
Also, I'd assume that Maemo explictily ships/supports a much smaller set of software than Fedora, so they can explicitly list all their dependencies that may need fixed.
Actually I would say more Qt apps are in Maemo repositories than in Fedora ones, that's really great, people are moving away from Gtk and switching to Qt! I hope we will see it in Fedora too soon, with more and more Qt-based apps!
Back to the topics.
Gnome Desktop spin is Qt free (unfortunately), same applies to other spins except Plasma Desktop spin. And there, we are pretty sure everything works as expected. So this change does not affect any default offering we explicitely ship/support.
And do not read this paragraph ;-)
You know I'm not a fan of banning updates but I'm really very conservative and very carefull about consequences. It's all about setting process to prepare high quality updates that makes user experience actually better. This involves communication (between developers, users, upstreams), testing, decision making etc. It's not easy task but that's our work. And I agree, we (as Fedora) failed here not a long time ago.
Jaroslav
Bill
Jaroslav Reznik wrote:
Actually I would say more Qt apps are in Maemo repositories than in Fedora ones, that's really great, people are moving away from Gtk and switching to Qt! I hope we will see it in Fedora too soon, with more and more Qt-based apps!
I'm sorry, but I do not wish to write in C++ while all my libraries I use are written C and I feel most comfortable writing in C. GTK+ and Qt can live happily *together*. There is no logical reason one needs to obsolete the other. Anything Qt can do I can do in GTK+. (Yes, this is an open invitation off-list to ask me.)
Back to the topics.
Yes, please.