In D18466#402362, @niccolove wrote:

This solution is Qt <5.12 friendly

Let's make public slot approach.

VersionControlObserver is exported class changing its constructor type parameter can brake ABI, but does it user of ? In other hand addition of public slot is ABI comfortable, so another approach can be better solution.

+1

So just don't call slotFilePrint connect loadCompleted to slotFilePrint. When openDocumentInternal downloads the document it fires signal of finish, just connect to slot of opened it.

I investigate on testing this patch, it does not fix systray issue, to discard it?

In D18466#399471, @danders wrote:

Probably we need to start printing in a slot connected to KoMainWindow::loadCompleted() instead.
And to be safe, we also need to handle loadCanceled().

@alexeymin, can you try this patch as well, it's pretty annoying to have 5+ versioned tabs, error in one to be shown in others.

In D18289#397527, @alexeymin wrote:

branches is a QStringList, so, we could also check that currentBranchIndex < branches.size() in condition, if it is all about bounds checking.

+1

ping, issue is real and annoying.

ping

+1 it looks good to me.

In D17640#393688, @elvisangelaccio wrote:

Sorry but I can't reproduce the bug you put in the commit message.

I tried to do a git checkout on a dirty repo, but the error message only shows up in the versioned tab (not in all tabs).

Searchbar propagates events even it's not visible, you can add check for visibility in focusin event.

Now in multi-thread environment will have problems especially on actualTheme(), no?

If you mean in 4.9 - no, i just not realize that.

Apparently it has a problem, if 2 or more versioned tabs are open (in distinct git directories) switching between does not update status/actions. It looks like in dolphin-plugin is used static object as well. So after all we don't want error in one tab to be multiplied to all opened, i've seen it often, since i'm using plugin daily, so easy fix is to update directory on tab activated, since observer has only private slots i'm using old syntax did you have better idea?

I want to have this in 18.12.1, i really not prefer to have custom changes against upstream.

The problem is too specific, it happen when dpms is activated after dimm display is setted screen to 0, if dimm option is not checked it works correct to me.

Video can't handle it. (i've think about screencast but can be recorded by a phone)

Does someone test it, @thsurrel ?

Ping, we can remove parent but i don't see advantage of this.

Ping, now on Alt+1 or right click ->project -> lookup does not show correct tab

@elvisangelaccio i want to go ahead with this patch.

If someone steal events it's not problem of the patch, similar patch as D15612 should be applied for that view.

Use parent in createInstance

As i mention before, that's not a patch issues.

@ngraham, you can test it.

ping

Can you push it, i've some problems - the change is small, thank you.

Remove locking inside media object
Still has a deadlock inside gstreamer

Thread 1 "dragon" received signal SIGINT, Interrupt.
0x00007ffff11f541c in __lll_lock_wait () from /lib/libpthread.so.0
(gdb) bt
#0  0x00007ffff11f541c in __lll_lock_wait () at /lib/libpthread.so.0
#1  0x00007ffff11eeb96 in pthread_mutex_lock () at /lib/libpthread.so.0
#2  0x00007fffc88d8001 in  () at /usr/lib/gstreamer-1.0/libgstplayback.so
#3  0x00007fffdb6f2d07 in gst_element_query () at /usr/lib/libgstreamer-1.0.so.0
#4  0x00007fffdb74a6ea in gst_element_query_position () at /usr/lib/libgstreamer-1.0.so.0
#5  0x00007fffdba41034 in Phonon::Gstreamer::Pipeline::position() const () at /usr/lib/qt5/plugins/phonon4qt5_backend/phonon_gstreamer.so
#6  0x00007fffdba32b5e in Phonon::Gstreamer::MediaObject::emitTick() [clone .part.17] () at /usr/lib/qt5/plugins/phonon4qt5_backend/phonon_gstreamer.so
#7  0x00007fffdba35c4f in Phonon::Gstreamer::MediaObject::qt_static_metacall(QObject*, QMetaObject::C--Type <RET> for more, q to quit, c to continue with all, int, void**) () at /usr/lib/qt5/plugins/phonon4qt5_backend/phonon_gstreamer.so
#8  0x00007ffff3b37e22 in QObject::event(QEvent*) () at /usr/lib/libQt5Core.so.5
#9  0x00007ffff4589e3c in QApplicationPrivate::notify_helper(QObject*, QEvent*) () at /usr/lib/libQt5Widgets.so.5
#10 0x00007ffff4590f5f in QApplication::notify(QObject*, QEvent*) () at /usr/lib/libQt5Widgets.so.5
#11 0x00007ffff3b098f8 in QCoreApplication::notifyInternal2(QObject*, QEvent*) () at /usr/lib/libQt5Core.so.5
#12 0x00007ffff3b0c295 in QCoreApplicationPrivate::sendPostedEvents(QObject*, int, QThreadData*) () at /usr/lib/libQt5Core.so.5
#13 0x00007ffff3b61313 in  () at /usr/lib/libQt5Core.so.5
#14 0x00007fffec5002f7 in g_main_context_dispatch () at /usr/lib/libglib-2.0.so.0
#15 0x00007fffec500508 in  () at /usr/lib/libglib-2.0.so.0
#16 0x00007fffec50058c in g_main_context_iteration () at /usr/lib/libglib-2.0.so.0
#17 0x00007ffff3b609af in QEventDispatcherGlib::processEvents(QFlags<QEventLoop::ProcessEventsFlag>)() at /usr/lib/libQt5Core.so.5
#18 0x00007fffe5f1c3b1 in  () at /usr/lib/qt5/plugins/platforms/../../../libQt5XcbQpa.so.5
#19 0x00007ffff3b081da in QEventLoop::exec(QFlags<QEventLoop::ProcessEventsFlag>) () at /usr/lib/libQt5Core.so.5
#20 0x00007ffff3b10cd0 in QCoreApplication::exec() () at /usr/lib/libQt5Core.so.5
#21 0x0000000000438514 in main ()

In facts that's 2 crashes, first one in ascent, second one misfit.

@danders you can commandeer this revision to update diff.

Yep, it's look like same approach to mine, did you try mine or it's not correct in all cases?

@broulik, @thsurrel any other ideas or to commit it?

Still has a problem, after aboutToFinish is emitted and queue is not empty, after setting new source paint of window freeze.

@danders that was one of my previous approach. Let @boemann tolds us which is correct one.

I saw another issue, when we waiting new source (MediaObject::setNextSource) it set m_pipeline to new source but since we have only one pipeline gstreamer starts notify us for states of new source which is paused state, that reflect to emit signal of state changed to paused. I don't know how pipeline more that one source ever works on gstreamer. I'll update patch.

Yeah, i know that, but i have a reason to do it - no cost :) Basically code base is same, so limitation is virtual, about me.

Adding context to diff

https://cgit.kde.org/dragon.git/commit/?h=Applications/18.12&id=addaf47d1af218a46777f740bdc9d579a16bdf1f

Also fix max comic strips limit