See attached. Reproducible: Always
Created attachment 386964 [details, diff] qt5.patch Patch sent to upstream mailing list.
I think I see a minor simplification of that patch: @@ -18,7 +18,11 @@ ***/ #include <sys/time.h> -#ifdef QT4 +#if #defined(QT5) +#include <QSocketNotifier> +#include <QObject> +#include <QTimer> +#elif defined(QT4) #include <Qt/qsocketnotifier.h> #include <Qt/qobject.h> #include <Qt/qtimer.h> AFAICT, #if #defined(QT5) || defined(QT4) #include <qsocketnotifier.h> #include <qobject.h> #include <qtimer.h> #endif should be correct for both.
You're right. Updating patch.
Created attachment 386968 [details, diff] qt5-r2.patch
(In reply to Michael Jones from comment #4) > Created attachment 386968 [details, diff] [details, diff] > qt5-r2.patch Ping me back when upstream accepts this and I'll back port it.
Posted to Avahi mailing list: http://lists.freedesktop.org/archives/avahi/2014-October/002305.html
https://github.com/lathiat/avahi/pull/19
(In reply to Michael Jones from comment #7) > https://github.com/lathiat/avahi/pull/19 qt5 support wasn't there as far as I can see in 0.6.32_rc which I just added to the tree, so I assume your pull/19 wasn't accepted yet? If its in the pipeline upstream, can't I just close this bug RESOLVED UPSTREAM and wait it trickles down to us?
You can, if you prefer to do it that way. You'll probably need to make changes to the ebuild to accommodate qt5. I'll post my hacked up ebuild for you to compare against. (Not that my ebuild skills are really impressive or anything.)
Created attachment 414662 [details] version of avahi ebuild with concessions to qt5.
And no, I don't think the pull request has been integrated yet. I suppose that means it'll be included in a later release.
(In reply to Michael Jones from comment #11) > And no, I don't think the pull request has been integrated yet. I suppose > that means it'll be included in a later release. no, they may just be reviewing it, i'm not sure. ping me back when its merged.
Easy Fix: Add export QT_SELECT=4 before the pkg_pretend() phase, then in the post_install() or at the end of one of the src_install phase re-export QT_SELECT=5, until this package patch is accepted upstraem
(In reply to Carter Young from comment #13) > Easy Fix: > > Add export QT_SELECT=4 before the pkg_pretend() phase, then in the > post_install() or at the end of one of the src_install phase re-export > QT_SELECT=5, until this package patch is accepted upstraem Just tested in my Local overlay, and this works. It even works when you don't reexport, in other words: > Easy Fix: > > Add export QT_SELECT=4 before the pkg_pretend() phase. is all that's needed. qmake -v still reports: QMake version 3.0 Using Qt version 5.6.1 in /usr/lib64 when properly symlinked using qtchooser.
Created attachment 445054 [details] Version -r1 of unpatched ebuild Attached version of original ebuild containing export
(In reply to Carter Young from comment #13) > Easy Fix: > > Add export QT_SELECT=4 before the pkg_pretend() phase, then in the > post_install() or at the end of one of the src_install phase re-export > QT_SELECT=5, until this package patch is accepted upstraem I'm confused here. Doesn't that cause this package to build against qt4? Anyhow, I'm tired of waiting for upstream. Two years is absurd. Can people just give Michael's patch a test and I'll just commit that. I'm sorry I waited so long on this but I just assumed upstream would act. It doesn't look like they are interested. I don't like carrying patches downstream but I'll do it if I have to.
(In reply to Anthony Basile from comment #16) > I'm confused here. Doesn't that cause this package to build against qt4? > > Anyhow, I'm tired of waiting for upstream. Two years is absurd. Can people > just give Michael's patch a test and I'll just commit that. I'm sorry I > waited so long on this but I just assumed upstream would act. It doesn't > look like they are interested. I don't like carrying patches downstream but > I'll do it if I have to. QT5 is only stable on the plasma profile, and there is no stable GCC5 compiler candidate in the portage tree. I'm not using the plasma profile, but I am using the GCC5 compiler, and as such, have qt-chooser installed. Packages like this are the reason qt-chooser is there. I'm not opposed to Micheal's patch, but I am opposed to early adoption of a patch for GCC5 support, when there is no stable GCC5 candidate in the tree.
Created attachment 500330 [details] avahi-0.7.ebuild with qt5 changes Attached new avahi-0.7.ebuild with qt5 changes
Created attachment 500332 [details, diff] avahi-0.7-qt5.patch Essentially identical to my previous patch. Major difference is that I force --std=gnu++11. Don't know why the compiler is defaulting to older now. Feel free to edit as desired.
Upstream seems uninterested in merging this patch. What's going to happen to avahi's qt integration when qt4 is removed from the tree?
@Anthony, Michael's patch works fine for me and also seems to solve bug 587830. Please, let's use this to get rid of USE=qt4 in here. Thanks Michael for your work!
Created attachment 508198 [details, diff] net-dns/avahi-0.7-r1.ebuild.diff
Created attachment 515442 [details, diff] net-dns/avahi-0.7-r1.ebuild.diff fix multilib-strict error
doesn't depend on qt5-multilib anymore.
(In reply to Andreas Sturmlechner from comment #24) > doesn't depend on qt5-multilib anymore. we should revbump this commit.
(In reply to Michael Jones from comment #20) > Upstream seems uninterested in merging this patch. > > What's going to happen to avahi's qt integration when qt4 is removed from > the tree? I pushed this patch against my better judgment. When upstream becomes uninterested in a patch I tend to shy away. Its not a question of correctness so much as who will maintain this going forward. I'll make an effort because there is value to qt-avahi integration, but if it becomes onerous, I'll drop the integration in future versions.
Let's keep this open until the old versions are gone.
The bug has been closed via the following commit(s): https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=5d34efe6e04c0eb4f83e66657805e4deb3298cb6 commit 5d34efe6e04c0eb4f83e66657805e4deb3298cb6 Author: Andreas Sturmlechner <asturm@gentoo.org> AuthorDate: 2018-05-27 08:16:21 +0000 Commit: Andreas Sturmlechner <asturm@gentoo.org> CommitDate: 2018-05-27 08:16:57 +0000 net-dns/avahi: Drop USE=qt4 in 0.6.32 hppa special edition Closes: https://bugs.gentoo.org/525884 Package-Manager: Portage-2.3.40, Repoman-2.3.9 net-dns/avahi/avahi-0.6.32.ebuild | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-)
