Commit by ossi :: r1254982 client/trunk/icecream/client/icecc-create-env: (link)

fix gcc 4.6 support for debian, take 2

the fix for debian (and derivates) is mutually exclusive with
everythinge else, so support both variants.

Commit by ossi :: r1233883 client/trunk/icecream/client/icecc-create-env: (link)

actually fix gcc 4.6 support

use the correct -print-foo option

Commit by ossi :: r1231874 kscreensaver/trunk/KDE/kdeartwork/kscreensaver/kdesavers/ (3 files): (link)

slideshow: add option to show full file path

based on patch by Robert Cheramy <robert at cheramy dot net>

FEATURE: 259489
FIXED-IN: 4.7

Commit by ossi :: r1230772 release/trunk/kde-common/release/anon: (link)

extra pedantry: fix find command

otherwise the -type d applies only to .svn, but not to .git

CCMAIL: mueller at kde dot org

Commit by ossi :: r1223891 kscreensaver/trunk/KDE/kdeartwork/kscreensaver/kdesavers/Flux.cpp: (link)

add missing glu.h include. fixes bug 261378

Commit by ossi :: r1223890 kscreensaver/branches/KDE/4.6/kdeartwork/kscreensaver/kdesavers/Flux.cpp: (link)

add missing glu.h include

BUG: 261378
FIXED-IN: 4.6.2

Commit by ossi :: r1196806 workspace/trunk/KDE/kdebase/workspace/kdm/kfrontend/kdmctl.c: (link)

forwardport: make kdmctl reliable

of course it's a bit naive to think that kdm's reply will arrive in one
piece, given that it writes chunkwise. fixes bug 256392

Commit by ossi :: r1196805 workspace/trunk/KDE/kdebase/workspace/kdm/backend/client.c: (link)

forwardport: fix crash on automatic re-login after automatic login

see https://bugzilla.redhat.com/show_bug.cgi?id=649345

Commit by ossi :: r1196804 workspace/branches/KDE/4.5/kdebase/workspace/kdm/kfrontend/kdmctl.c: (link)

make kdmctl reliable

of course it's a bit naive to think that kdm's reply will arrive in one
piece, given that it writes chunkwise.

BUG: 256392
FIXED-IN: 4.5.4

Commit by ossi :: r1196803 workspace/branches/KDE/4.5/kdebase/workspace/kdm/backend/client.c: (link)

fix crash on automatic re-login after automatic login

see https://bugzilla.redhat.com/show_bug.cgi?id=649345

CCMAIL: Than Ngo <than at redhat dot com>

Commit by ossi :: r1196752 kdecore/trunk/KDE/kdelibs/kdecore/ (3 files in 2 dirs): (link)

remove kcmdwrapper hack

with the introduction of QProcess::setNativeArguments() in Qt 4.7, this
somewhat arcane indirection meant for bypassing QProcess' argument
quoting became obsolete. advantages are a slight performance gain, sane
KProcess::kill() behavior, and less bizarreness in general.

Commit by ossi :: r1196658 runtime/trunk/KDE/ (5 files in 2 dirs): (link)

support control chars in commands executed via kdesu

FEATURE: 198967
FIXED-IN: 4.6

Commit by ossi :: r1196651 kdesu/trunk/KDE/kdelibs/kdesu/process.cpp: (link)

optimize a bit

use a bigger buffer. this may matter if somebody uses the kdesu -t
option to pipe bigger amounts of data somewhere.

Commit by ossi :: r1196650 kdesu/trunk/KDE/kdelibs/kdesu/kdesu_stub.c: (link)

forwardport: reset DBUS_SESSION_BUS_ADDRESS

this avoids that the su'd process tries to connect to a session bus
which rejects it immediately.

fixes bug 217620

Commit by ossi :: r1196647 kdesu/branches/KDE/4.5/kdelibs/kdesu/kdesu_stub.c: (link)

reset DBUS_SESSION_BUS_ADDRESS

this avoids that the su'd process tries to connect to a session bus
which rejects it immediately.

BUG: 217620
FIXED-IN: 4.5.4

Commit by ossi :: r1196645 runtime/trunk/KDE/kdebase/runtime/kdesu/kdesu/kdesu.cpp: (link)

change command line interpretation

when the -c option is not used, fully quote even the first argument.
that's behavior incompatible, but it's way more expected.

Commit by ossi :: r1196644 runtime/trunk/KDE/kdebase/runtime/kdesu/kdesud/kdesud.cpp: (link)

use _exit() to quit the daemon parent

Commit by ossi :: r1196642 runtime/trunk/KDE/kdebase/runtime/kdesu/kdesud/kdesud.cpp: (link)

forwardport: make socket handling more reliable

make the socket non-blocking, so things don't go awry when a client
freezes. the socket reading will be able to deal with that. the
socket writing function will simply ignore errors - that's not a
problem, as the socket's internal buffer is big enough for our
payloads, so blocking would indicate a serious problem anyway.

set the accept() queue length to 10. no reason to limit it to one.

speculative fix for bug 208979

Commit by ossi :: r1196641 runtime/trunk/KDE/kdebase/runtime/kdesu/kdesud/handler.cpp: (link)

forwardport: fix up buffer handling

eliminate separate temporary buffer for simplicity, and make the
(somewhat pointless) buffer erasing code actually work.

Commit by ossi :: r1196639 runtime/branches/KDE/4.5/kdebase/runtime/kdesu/kdesud/kdesud.cpp: (link)

make socket handling more reliable

make the socket non-blocking, so things don't go awry when a client
freezes. the socket reading will be able to deal with that. the
socket writing function will simply ignore errors - that's not a
problem, as the socket's internal buffer is big enough for our
payloads, so blocking would indicate a serious problem anyway.

set the accept() queue length to 10. no reason to limit it to one.

speculative fix for
CCBUG: 208979
FIXED-IN: 4.5.4