Bug 335161

Eh, what do you mean? I can not reproduce.. screenshot maybe?

Created attachment 86775 [details]
Screenshot

Notice the EEEEEEEEE. Is it from pt_BR translation?

it looks like translation problem..

There's no EEEEE in pt_BR translations (current master)

void DesktopNamesWidget::setDefaultName(int desktop)
{
    if ((desktop < 1) || (desktop > m_maxDesktops))
        return;
    QString name = m_desktopConfig->cachedDesktopName(desktop);
    if (name.isEmpty())
        name = i18n("Desktop %1", desktop);
    m_nameInputs[ desktop -1 ]->setText(name);
}

Luiz, can you confirm if the bug is still there and which translations you are using.

Re-assigning to kwin which is where the KCM is.

What's the output of

   kreadconfig --file kwinrc --group Desktops --key Name_1

$ kreadconfig --file kwinrc --group Desktops --key Name_1

Did not work (there is no kreadconfig for kf5 in opensuse packages. It might tried to read kde4 files)

However, I got the kwinrc file directly.

cat ~/.config/kwinrc
[Compositing]
OpenGLIsUnsafe=false

[Desktops]
Name_1=EEEEEEEEE
Number=1
Rows=1

[Effect-Zoom]
InitialZoom=1

[Plugins]
cubeslideEnabled=false
desktopchangeosdEnabled=false
kwin4_effect_fadedesktopEnabled=false
slideEnabled=true

[Script-desktopchangeosd]
PopupHideDelay=1000
TextOnly=false

[Windows]
RollOverDesktops=true

 Is it maybe a downstream problem?
I looked for this EEEEE and found nothing:

$ rpm -qa --queryformat "%{Name}:%{Vendor}\n" | grep obs://build.opensuse.org/KDE:Unstable | cut -d: -f1 | xargs rpm -ql | sed 's/.*/"&"/' | xargs strings | grep EEEEEEEEE
FIXXXXXMMEEEEEEEEEEEEE
N3WTF10RefCountedIN7WebCore7SVGListINS_6RefPtrINS1_14SVGPODListItemINS1_12SVGTransformEEEEEEEEE
N3WTF10RefCountedIN7WebCore7SVGListINS_6RefPtrINS1_14SVGPODListItemINS1_10FloatPointEEEEEEEEE
N3WTF10RefCountedIN7WebCore7SVGListINS_6RefPtrINS1_14SVGPODListItemINS1_9SVGLengthEEEEEEEEE
N3WTF10RefCountedIN7WebCore7SVGListINS_6RefPtrINS1_14SVGPODListItemINS1_12SVGTransformEEEEEEEEE
N3WTF10RefCountedIN7WebCore7SVGListINS_6RefPtrINS1_14SVGPODListItemINS1_10FloatPointEEEEEEEEE
N3WTF10RefCountedIN7WebCore7SVGListINS_6RefPtrINS1_14SVGPODListItemINS1_9SVGLengthEEEEEEEEE
strings: /var/lib/NetworkManager: Permissão negada
strings: /var/log/NetworkManager: Permissão negada

And also nothing on /etc

The only thing I know is that if I remove all .config contents, kde generate a new config with the EEEEEEEEE.

it's kreadconfig5 with kf5.

KWin saves the configuration of the desktop names before shutting down. As you have a desktop called EEEEEEEEE, it gets saved. You have to go to the desktop settings KCM and click defaults and apply them.

I don't know where the strange name came from, but not from KWin, thus setting to worksforme.

*** Bug 335436 has been marked as a duplicate of this bug. ***

Reopen? There is another report (Bug 335436)

@romario
what does your locale happen to be...?

(In reply to comment #10)
> @romario
> what does your locale happen to be...?
pt_BR

That's what I call a pattern - can you try pt?
There's sth. interesting:

pt is "Ecrã %1" whilte pt_BR id "Área de trabalho %1"

(In reply to comment #12)
> That's what I call a pattern - can you try pt?
Well, I can't find the language KCM in systemsettings. Where do I change the system language?

> There's sth. interesting:
> 
> pt is "Ecrã %1" whilte pt_BR id "Área de trabalho %1"
OT: The pt_BR translation literally translates to "Working area", while the pt_PT translation -- I think -- is an adaptation from the French "écran", which translates to "Screen".

Created attachment 86883 [details]
Localization section in systemsettings

This is all I have in the localization section in systemsettings. There's no language selection KCM.

Created attachment 86890 [details]
Screenshot

Correct translation (Área de trabalho) with Project Neon5.

should be reassigned IMO. have the EEEEEEE with no lang selected, and with also croatian lang.

Created attachment 86916 [details]
screenshot

where is that coming from? general serach of lxr doesn't find it, so I rather 
doubt it's in our sources.

@Hrvoje: any chance to figure out which package contains the string?

Interesting, it's not "EEEEEEEEEEEEEEEEo 1" but just "EEEEEEEEE" for you.
Since it should be "Área de trabalho %1" for pt_BR, that means the first 16 bytes are overridden with "E", ie. sth.'s bleeding in here.

What about "Desktop 2"?
"EEEEEEEEE", resp. "EEEEEEEEEEEEEEEEo 2" or some realy word?

(In reply to comment #18)
> @Hrvoje: any chance to figure out which package contains the string?

will try

(In reply to comment #19)
> What about "Desktop 2"?
> "EEEEEEEEE", resp. "EEEEEEEEEEEEEEEEo 2" or some realy word?

as per screenshot, every additional desktop created has 'Desktop $number'

I don't think this string will be anywhere - it's rather some data bleeding into the read config value.

virtualdesktop.cpp:334

- QString s = group.readEntry(QStringLiteral("Name_%1").arg(i), i18n("Desktop %1", i));
+ QString s = i18n("Desktop %1", i);
+ qDebug() << "i18n: << s;
+ s = group.readEntry(QStringLiteral("Name_%1").arg(i), s); 
+ qDebug() << "value: << s << s.toUtf8() << s.toUtf8().data();

That junk should either come from i18n(), group.readEntry() or is ultimately applied by
m_rootInfo->setDesktopName(i, s.toUtf8().data());

If the above debug outs are sane, it's writing the data onto the root window.

additional observation:
if i remove [Desktops] group (while e.g. kwin4 is running),  restart kwin5, and start kscmshell5 desktop,  it will just write in
Number=1
in kwinrc (and kcm correctly has Desktop 1)

(In reply to comment #21)
> If the above debug outs are sane, it's writing the data onto the root window.

i18n: "Desktop 1"
value: "Desktop 1" "Desktop 1" Desktop 1

as it 'works elsewhere', attaching compiler command line for kdeinit_kwin target:

/usr/bin/c++   -DKCOREADDONS_LIB -DKDE_DEFAULT_DEBUG_AREA=1212 -DQT_CONCURRENT_LIB -DQT_CORE_LIB -DQT_DBUS_LIB -DQT_DISABLE_DEPRECATED_BEFORE=0 -DQT_GUI_LIB -DQT_NETWORK_LIB -DQT_NO_DEBUG -DQT_QML_LIB -DQT_QUICK_LIB -DQT_SCRIPT_LIB -DQT_WIDGETS_LIB -DQT_X11EXTRAS_LIB -DQT_XML_LIB -D_GNU_SOURCE -D_LARGEFILE64_SOURCE -D_XOPEN_SOURCE=500 -Dkdeinit_kwin_EXPORTS -fmessage-length=0 -grecord-gcc-switches -O2 -Wall -D_FORTIFY_SOURCE=2 -fstack-protector -funwind-tables -fasynchronous-unwind-tables -g -std=c++0x -fno-exceptions -Wall -Wextra -Wcast-align -Wchar-subscripts -Wformat-security -Wno-long-long -Wpointer-arith -Wundef -Wnon-virtual-dtor -Woverloaded-virtual -Werror=return-type -O2 -g -DNDEBUG -fPIC -fvisibility=hidden -fvisibility-inlines-hidden -I....

packages are built in release mode (but OBS adding -g makes it ~relwithdebinfo)

I noticed that if I add more desktops in systemsettings, all comes with the correct translation. The problem is only when profile is generated.

can someone try to inject a flush before deleting the string?

kwindowsystem.git, netwm.cpp:948

void NETRootInfo::setDesktopName(int desktop, const char *desktopName)
{
...
xcb_change_property(p->conn, XCB_PROP_MODE_REPLACE, p->root, net_desktop_names,
                        UTF8_STRING, 8, proplen, (const void *) prop);
+ xcb_flush(p->conn); 
delete [] prop;
}

no change

Hmmm ... on read?
What's the output of "xprop -root _NET_DESKTOP_NAMES"?

also good,
_NET_DESKTOP_NAMES(UTF8_STRING) = "Desktop 1"

eeewwww....

kwindowsystem.git,  netwm.cpp:358
        if (data) {
            const QByteArray ba = QByteArray::fromRawData(data, data[len - 1] ? len : len - 1);
            list = ba.split('\0');
+            if (list.count() == 1) {
+               QByteArray deepCopy(list.at(1).constData());
+                list.clear();
+                list << deepCopy;
+            }
        }

[KCrash Handler]
#5  data (this=0xbabababababababa) at /usr/include/qt5/QtCore/qarraydata.h:63
#6  data (this=0xbabababababababa) at /usr/include/qt5/QtCore/qarraydata.h:211
#7  constData (this=0x2234f18) at /usr/include/qt5/QtCore/qbytearray.h:451
#8  get_stringlist_reply (c=<optimized out>, cookie=..., cookie@entry=..., type=<optimized out>) at /home/hrvoje/Src/local/kwindowsystem/src/netwm.cpp:362
#9  0x00007f9127181618 in NETRootInfo::update (this=<optimized out>, properties=..., properties2=...) at /home/hrvoje/Src/local/kwindowsystem/src/netwm.cpp:2325
#10 0x00007f912718332c in NETRootInfo::NETRootInfo (this=0x2238b90, connection=<optimized out>, supportWindow=52428807, wmName=0x7f912a63e9ff "KWin", properties=..., windowTypes=..., states=..., properties2=..., actions=..., screen=0, doActivate=true) at /home/hrvoje/Src/local/kwindowsystem/src/netwm.cpp:698
#11 0x00007f912a4f38c9 in RootInfo (scr=0, actions=..., properties2=..., states=..., types=..., properties=..., name=0x7f912a63e9ff "KWin", w=52428807, this=0x2238b90) at /usr/src/debug/kwin-4.95.0~20140530~0bf9ed5/netinfo.cpp:143
#12 KWin::RootInfo::create () at /usr/src/debug/kwin-4.95.0~20140530~0bf9ed5/netinfo.cpp:128
#13 0x00007f912a4cf787 in KWin::Workspace::init (this=this@entry=0x216cfa0) at /usr/src/debug/kwin-4.95.0~20140530~0bf9ed5/workspace.cpp:264
#14 0x00007f912a4d13b6 in KWin::Workspace::Workspace (this=0x216cfa0, restore=<optimized out>) at /usr/src/debug/kwin-4.95.0~20140530~0bf9ed5/workspace.cpp:232
#15 0x00007f912a5016cb in operator() (__closure=<synthetic pointer>) at /usr/src/debug/kwin-4.95.0~20140530~0bf9ed5/main.cpp:271
#16 call (arg=<optimized out>, f=...) at /usr/include/qt5/QtCore/qobjectdefs_impl.h:502
#17 call<QtPrivate::List<>, void> (arg=<optimized out>, f=...) at /usr/include/qt5/QtCore/qobjectdefs_impl.h:559
#18 QtPrivate::QFunctorSlotObject<KWin::Application::start()::__lambda8, 0, QtPrivate::List<>, void>::impl(int, QtPrivate::QSlotObjectBase *, QObject *, void **, bool *) (which=<optimized out>, this_=<optimized out>, r=<optimized out>, a=<optimized out>, ret=<optimized out>) at /usr/include/qt5/QtCore/qobject_impl.h:200
#19 0x00007f9122061283 in QMetaObject::activate(QObject*, int, int, void**) () from /usr/lib64/libQt5Core.so.5
#20 0x00007f9127163952 in KSelectionOwner::Private::claimSucceeded (this=0x20b6760) at /home/hrvoje/Src/local/kwindowsystem/src/kselectionowner.cpp:180
#21 0x00007f9127164169 in KSelectionOwner::filterEvent (this=0x20b6540, ev_P=<optimized out>) at /home/hrvoje/Src/local/kwindowsystem/src/kselectionowner.cpp:398
#22 0x00007f912202fbe0 in QAbstractEventDispatcher::filterNativeEvent(QByteArray const&, void*, long*) () from /usr/lib64/libQt5Core.so.5
#23 0x00007f91144eb078 in QXcbConnection::handleXcbEvent(xcb_generic_event_t*) () from /usr/lib64/qt5/plugins/platforms/libqxcb.so
#24 0x00007f91144ec3eb in QXcbConnection::processXcbEvents() () from /usr/lib64/qt5/plugins/platforms/libqxcb.so
#25 0x00007f9122062406 in QObject::event(QEvent*) () from /usr/lib64/libQt5Core.so.5
#26 0x00007f9122cd2ecc in QApplicationPrivate::notify_helper(QObject*, QEvent*) () from /usr/lib64/libQt5Widgets.so.5
#27 0x00007f9122cd7c46 in QApplication::notify(QObject*, QEvent*) () from /usr/lib64/libQt5Widgets.so.5
#28 0x00007f9122032825 in QCoreApplication::notifyInternal(QObject*, QEvent*) () from /usr/lib64/libQt5Core.so.5
#29 0x00007f9122034677 in QCoreApplicationPrivate::sendPostedEvents(QObject*, int, QThreadData*) () from /usr/lib64/libQt5Core.so.5
#30 0x00007f9122087694 in QEventDispatcherUNIX::processEvents(QFlags<QEventLoop::ProcessEventsFlag>) () from /usr/lib64/libQt5Core.so.5
#31 0x00007f911453b0fd in QUnixEventDispatcherQPA::processEvents(QFlags<QEventLoop::ProcessEventsFlag>) () from /usr/lib64/qt5/plugins/platforms/libqxcb.so
#32 0x00007f912203073b in QEventLoop::exec(QFlags<QEventLoop::ProcessEventsFlag>) () from /usr/lib64/libQt5Core.so.5
#33 0x00007f9122037d96 in QCoreApplication::exec() () from /usr/lib64/libQt5Core.so.5
#34 0x00007f912a503a95 in kdemain (argc=3, argv=0x7fff553db588) at /usr/src/debug/kwin-4.95.0~20140530~0bf9ed5/main.cpp:554
#35 0x00007f912a0f2b05 in __libc_start_main () from /lib64/libc.so.6
#36 0x00000000004008de in _start () at ../sysdeps/x86_64/start.S:122

Sorry, i'm stupid - ".at(0)" will certainly work MUCH better ;-)

not only that, but i have now a recognizable name - 'Desktop 1' ;-)

Ok, let's fix the bug status then - this is not "minor" at all.

----------

The problem here is that QByteArray::fromRawData() is "implicitly shared" with the provided POD and that QByteArray::split() will basically just "return QList<QByteArray>() << *this;" if the BA doesn't contain the split parameter ("\0"), so you get a QBA List that only contains the QBA with the shared POD ... which is freed before the return.
So we need an explicit deep copy of the QBA in the list instead.

Some compiler/parameter combo apparently does not only free() but also EEEEEEEEEEEEEEEE the first 16 bytes of the freed data, but the data is in general invalid and can take any content any time.

(In reply to comment #34)
> Some compiler/parameter combo apparently does not only free() but also
> EEEEEEEEEEEEEEEE the first 16 bytes of the freed data, but the data is in
> general invalid and can take any content any time.

tried building (w/o patch) kwindowsystem:
omitting any 'custom' flag
building in debug mode
-> EEEEEEEEEE

i wonder is it something in Qt setup then?

could be hardcoded/patched into glibc then - do we have any non-suse eeeeeeeeeets?

(just to cause no irritation in this thread:
bug is clearly in kwindowsystem. whatever whoever does on free'd memory is just fine)

one new note (maybe not unexpected);
if i add
[Desktops]
Number=2
to /etc/xdg/kwinrc, and test with a clean config, both desktops are correctly named

i can't reproduce anymore with current master + Qt 5.4

Thank you for the bug report.

As this report hasn't seen any changes in 5 years or more, we ask if you can please confirm that the issue still persists.

If this bug is no longer persisting or relevant please change the status to resolved.

Comment on attachment 86883 [details]
Localization section in systemsettings

sadsgrhtk,n

Created attachment 168749 [details]
aDSFDHNM

XAFD