Perform asynchronous connection tests before getting destinations
from CUPS server and before creating a subscription for CUPS
messages (#677081).

Matthias Clasen authored Apr 23, 2012 and Marek Kašík committed Jan 08, 2013

The start() function is running before gsd finishes its
session registration, so everything we do here is in the
critical path for session startup. Both cupsGetDests and
the subscription registration are potentially expensive,
and can very well be done later.

https://bugzilla.gnome.org/show_bug.cgi?id=674551

Matthias Clasen authored Apr 21, 2012 and Marek Kašík committed Jan 08, 2013

The proxy object was not used for anything, and it was creating
synchronously.

https://bugzilla.gnome.org/show_bug.cgi?id=674550

If the user has connected a device without the ability to get the EDID (e.g.
composite TV connection) then don't show a critical warning and fail to add
the device.

Resolves https://bugzilla.redhat.com/show_bug.cgi?id=749504

Antoine Jacoutot authored Dec 16, 2011 and Richard Hughes committed Mar 15, 2012

The /sys filesystem is only available on Linux, so there is no need to
throw lots of warnings about /sys/class/dmi/... not being available on
other OSes.
https://bugzilla.gnome.org/show_bug.cgi?id=666350

A lot of laptops have invalid EDIDs, but it doesn't matter as we prefer the
DMI data anyway.

It appears quite a lot of real-world hardware returns bad EDID data.

When an EDID is valid, but crap (for example in a cheap projector) we don't get
any EDID vendor, model or serial. Fall back to the connection name in this case,
rather than just registering an 'xrandr' device ID.

According to the colord device-and-profile-naming-spec.txt it should be
'xrandr-lvds1', not 'xrandr_lvds1'.

Do not try to access a device ID when the device does not exist.

Resolves: https://bugzilla.gnome.org/show_bug.cgi?id=670599

When there is no primary or laptop display, fall back to the first *connected*
display, not just the first display returned from xrandr.

Resolves: https://bugzilla.gnome.org/show_bug.cgi?id=670368

Note, in C is doesn't matter as they are both enums of the same value, but if
you copy this code and use it with g++ you get a warning. Use the correct type
in case gcc ever gets this strict too.

X sends a message to us to tell us when the accessx settings are
changed.  They can change for two reasons: because a client (like
ourselves) has changed them or because the user did some keystroke (like
pressing shift 5 times to enable stickykeys).

We want to update GSettings in response to user requests, but prevent
spurious updates caused by the X server notifying us of the updates that
we just did for ourselves, so check that the update is actually caused
by user input.

https://bugzilla.gnome.org/show_bug.cgi?id=668213

Alexandre Rostovtsev authored Jan 11, 2012 and Bastien Nocera committed Jan 12, 2012

Needed for gdk_x11_device_manager_lookup()

https://bugzilla.gnome.org/show_bug.cgi?id=667731

In GNOME 2.32 we used a heuristic to work out what output profile should be set
screen-wide.

When converting the color process to g-s-d for GNOME 3, we switched to using the
primary display as the key to which monitor is the most important. This works
for most people, but some video drivers do not set a primary display when using
'xrandr --auto' and hence do not get a screen atom set.

In this case, fall back to the old heuristic when there are no primary monitors
displayed. This fixes applications like GIMP that expect a screen atom to be set
to properly color manage the display.

NOTE: We can't change GnomeRR to always return a primary monitor using the
same heuristic, otherwise it stops the screen control center capplet from
assigning the primary device correctly.

Lars Uebernickel authored Dec 07, 2011 and Marek Kašík committed Dec 07, 2011

Fixes a crash that happens when receiving a d-bus call with an invalid
number of arguments.  The issue was first reported at lp #893676.  A
patch to CUPS which makes it stop sending those malformed signals has
been submitted to cups.org as STR #3984. (#665689)

Dominique Leuenberger authored Nov 26, 2011 and Richard Hughes committed Nov 26, 2011

The vendor is being read, it was just not added to the text.
Resolves: #664418

Signed-off-by: Richard Hughes <richard@hughsie.com>

If the user has connected a device without the ability to get the EDID (e.g.
composite TV connection) then don't show a critical warning and fail to add
the device.

Resolves https://bugzilla.redhat.com/show_bug.cgi?id=749504

0 != GDK_KEY_VoidSymbol

gdk_keyval_from_name() doesn't seem to work as it used to.

https://bugzilla.gnome.org/show_bug.cgi?id=642489

Free old_state_reasons and new_state_reasons properly.

Make notifications for printer-state-reasons resident and
set their timeout to 15s so that users don't miss them.
(rhbz #743944)

https://bugzilla.gnome.org/show_bug.cgi?id=663623

Peter Hutterer authored Nov 04, 2011 and Bastien Nocera committed Nov 08, 2011

XI2_Major/XI2_Minor are the _current_ version of the protocol as exported by
the inputproto header files. If gsd is compiled against new headers and runs
against a 2.0 server it reports "no XInput2 supported" otherwise.

Remove "/" from allowed characters so that models which contain
this character can be added successfully (#661774).

Christian Persch authored Nov 03, 2011 and Bastien Nocera committed Nov 03, 2011

==22370== 708 bytes in 12 blocks are definitely lost in loss record 11,519 of 11,921
==22370==    at 0x402AD89: malloc (vg_replace_malloc.c:236)
==22370==    by 0x4E68023: XGetFeedbackControl (XGetFCtl.c:132)
==22370==    by 0x7F5D1EA: set_motion (gsd-mouse-manager.c:451)
==22370==    by 0x7F5EC5C: set_mouse_settings (gsd-mouse-manager.c:894)
==22370==    by 0x7F5EF86: gsd_mouse_manager_idle_cb (gsd-mouse-manager.c:1097)

Bug #663239.

Don't show printer-state-reasons when a job trigerring the change has been
sent by another user.
Create a hash table of printers for jobs which are processed. Check this
hash table in callback of PrinterStateChanged signal. (rhbz #744144)

Christian Persch authored Nov 03, 2011 and Bastien Nocera committed Nov 03, 2011

==24418== 1 bytes in 1 blocks are definitely lost in loss record 32 of 10,009
==24418==    at 0x402AD89: malloc (vg_replace_malloc.c:236)
==24418==    by 0x4A62C3B: standard_malloc (gmem.c:88)
==24418==    by 0x4A631B0: g_malloc (gmem.c:164)
==24418==    by 0x4A7B3DB: g_strdup (gstrfuncs.c:100)
==24418==    by 0x4A952CB: g_variant_dup_string (gvariant.c:1355)
==24418==    by 0x490EDF0: g_settings_get_string (gsettings.c:1769)
==24418==    by 0x804B38A: got_session_proxy (main.c:209)

Bug #663239.

Christian Persch authored Nov 02, 2011 and Bastien Nocera committed Nov 03, 2011

==4196== 1,105 bytes in 1 blocks are definitely lost in loss record 11,035 of 11,286
==4196==    at 0x402AD89: malloc (vg_replace_malloc.c:236)
==4196==    by 0x4E6961A: XListInputDevices (XListDev.c:208)
==4196==    by 0x6EBEEF3: get_disabled_devices (gsd-input-helper.c:334)
==4196==    by 0x7F5F065: gsd_mouse_manager_idle_cb (gsd-mouse-manager.c:1108)

Bug #663239.

Christian Persch authored Nov 02, 2011 and Bastien Nocera committed Nov 03, 2011

g_settings_get_string() returns non-NULL strings.

Bug #663239.

Christian Persch authored Nov 02, 2011 and Bastien Nocera committed Nov 03, 2011

==24379== 88 bytes in 3 blocks are definitely lost in loss record 5,392 of 6,164
==24379==    at 0x402AD89: malloc (vg_replace_malloc.c:236)
==24379==    by 0x4A62C3B: standard_malloc (gmem.c:88)
==24379==    by 0x4A631B0: g_malloc (gmem.c:164)
==24379==    by 0x4A526D3: g_key_file_parse_value_as_string (gkeyfile.c:3673)
==24379==    by 0x4A543F8: g_key_file_get_string (gkeyfile.c:1438)
==24379==    by 0x7DBE1FB: get_gtk_modules_from_dir (gsd-xsettings-gtk.c:121)

==24379== 59 bytes in 3 blocks are definitely lost in loss record 5,034 of 6,164
==24379==    at 0x402AD89: malloc (vg_replace_malloc.c:236)
==24379==    by 0x4A62C3B: standard_malloc (gmem.c:88)
==24379==    by 0x4A631B0: g_malloc (gmem.c:164)
==24379==    by 0x4A526D3: g_key_file_parse_value_as_string (gkeyfile.c:3673)
==24379==    by 0x4A543F8: g_key_file_get_string (gkeyfile.c:1438)
==24379==    by 0x7DBE223: get_gtk_modules_from_dir (gsd-xsettings-gtk.c:122)

Bug #663239.

Christian Persch authored Nov 02, 2011 and Bastien Nocera committed Nov 03, 2011

==24379== 69 bytes in 4 blocks are definitely lost in loss record 5,253 of 6,164
==24379==    at 0x402AD89: malloc (vg_replace_malloc.c:236)
==24379==    by 0x4A62C3B: standard_malloc (gmem.c:88)
==24379==    by 0x4A631B0: g_malloc (gmem.c:164)
==24379==    by 0x4A526D3: g_key_file_parse_value_as_string (gkeyfile.c:3673)
==24379==    by 0x4A543F8: g_key_file_get_string (gkeyfile.c:1438)
==24379==    by 0x7DBE13E: get_gtk_modules_from_dir (gsd-xsettings-gtk.c:110)

Bug #663239.

Christian Persch authored Nov 02, 2011 and Bastien Nocera committed Nov 03, 2011

==24379== 9 bytes in 1 blocks are definitely lost in loss record 1,324 of 6,164
==24379==    at 0x402AD89: malloc (vg_replace_malloc.c:236)
==24379==    by 0x4A62C3B: standard_malloc (gmem.c:88)
==24379==    by 0x4A631B0: g_malloc (gmem.c:164)
==24379==    by 0x4A7B687: g_strconcat (gstrfuncs.c:325)
==24379==    by 0x7DBC77C: update_property (gsd-xsettings-manager.c:338)
==24379==    by 0x7DBCBE1: update_xft_settings (gsd-xsettings-manager.c:375)

Bug #663239.

Christian Persch authored Nov 02, 2011 and Bastien Nocera committed Nov 03, 2011

==24379== 4 bytes in 1 blocks are definitely lost in loss record 203 of 6,164
==24379==    at 0x402AD89: malloc (vg_replace_malloc.c:236)
==24379==    by 0x4A62C3B: standard_malloc (gmem.c:88)
==24379==    by 0x4A631B0: g_malloc (gmem.c:164)
==24379==    by 0x4A63522: g_malloc_n (gmem.c:381)
==24379==    by 0x4A95464: g_variant_dup_strv (gvariant.c:1471)
==24379==    by 0x490F110: g_settings_get_strv (gsettings.c:2041)
==24379==    by 0x7DBE38C: update_gtk_modules (gsd-xsettings-gtk.c:226)

==24379== 4 bytes in 1 blocks are definitely lost in loss record 204 of 6,164
==24379==    at 0x402AD89: malloc (vg_replace_malloc.c:236)
==24379==    by 0x4A62C3B: standard_malloc (gmem.c:88)
==24379==    by 0x4A631B0: g_malloc (gmem.c:164)
==24379==    by 0x4A63522: g_malloc_n (gmem.c:381)
==24379==    by 0x4A95464: g_variant_dup_strv (gvariant.c:1471)
==24379==    by 0x490F110: g_settings_get_strv (gsettings.c:2041)
==24379==    by 0x7DBE3A6: update_gtk_modules (gsd-xsettings-gtk.c:227)

Bug #663239.

Marien Zwart authored Nov 03, 2011 and Bastien Nocera committed Nov 03, 2011

gnome-settings-daemon draws a correct osd window when you hit a
media key, but does not redraw that window on volume changes afterwards,
unless a compositing manager is running (tested with xcompmgr).

It looks like the culprit is gsd_osd_window_update_and_hide, which
has a description of "Queues the @window for immediate drawing, and
queues a timer to hide the window." but only queues a redraw if the
window is composited.

Removing the check for compositedness makes it work without a
compositing manager.

https://bugzilla.gnome.org/show_bug.cgi?id=660990