* qt/Makefile.am: Patch by Sjoerd Simons. More .moc issues:
make/automake don't detect that we're talking about the same
.lo file if I specify the full path to the source files.
* bus/dbus-daemon.1.in: improve the language in a couple spots I noticed
* dbus/dbus-bus.c (internal_bus_get): in the error message if the
session bus variable is unset, suggest "man dbus-launch" and "man
dbus-daemon" to figure out how to fix the problem
* qt/qdbusinterface_p.h:
* qt/qdbusinterface.cpp: Use the standard
org.freedesktop.DBus.Method.NoReply annotation for the "async"
calls instead of creating one for us.
* qt/qdbusconnection_p.h:
* qt/qdbusintegrator.cpp: Remove debugging code.
* qt/qdbusintegrator.cpp:
* qt/qdbusmessage.cpp:
* qt/qdbusmessage_p.h:
* qt/qdbusmessage.h: Change the behaviour of automatic
reply-sending: now a reply is always sent, unless the caller
didn't request one or if the user slot has already sent one.
* configure.in: Patch from Debian packages by Sjoerd Simons
<sjoerd@debian.org> to add --with-qt-moc and --with-qt3-moc
arguments so it's possible to build both bindings in the
same tree.
* qt/Makefile.am: Fix truncated value so that make dist works.
* acinclude.m4, configure.in: Patch from Brad Hards
<bradh@frogmouth.net> to avoid warnings from autoconf 1.9 by
improving quoting, re-ordering a few checks, and a few other
aesthetic tidy-ups.
* dbus/dbus-message.c (dbus_message_iter_get_fixed_array):
Patch from Rob Taylor <rob.taylor@collabora.co.uk> to correct a bogus
assertion that the next element to read from the iter is fixed in
size. This is not the case when you are at the end of the iter,
because the next element type is INVALID.
* dbus/dbus-string.c (_dbus_string_init_const_len): Correct a
a bogus assert which means that you may not initialise a 0-length
string unless you provide a non-NULL pointer. This prevented
you from marshalling messages containing zero-length arrays in
some cases.
* glib/dbus-gvalue.c (demarshal_collection_array): Another patch
from Rob to correct bogus asserts when trying to demarshal an
array and get_fixed_array got you 0 elements. Append nothing to
the GArray in this case.
* test/glib/test-dbus-glib.c: Add a test case for round-tripping
an empty array via the glib bindings. Without all of the above
patches, this new test fails.
* glib/dbus-gmain.c: Make the previous commit compile.
* python/_dbus.py, python/matchrules.py: Patch from Ole Andre
Ravnaas <ole.andre.ravnaas@collabora.co.uk> to allow you to
specify sender_keyword="foo", path_keyword="bar" when adding
a signal listener, so that you can bind to signals generically
but still do something useful in your callback.
* python/dbus_bindings.pyx: Demarshal the byte type as unsigned
chars so that they're not cast to chars and made negative. Thanks
to Jakub Stachowski for reporting this and testing the fix.
* dbus/dbus-glib.h:
* glib/dbus-gmain.h:
(dbus_g_connection_open): new method for openning
a connection to an arbitrary address in the glib bindings
* ChangeLog: checkin last entry which doesn't seem to be commited
This is a major change: library is source- and binary-incompatible to
what it used to be.
All testcases are green, functionality is preserved.
It is not feature-complete. Development will continue in the branch in the
Subversion server for a while.
* glib/dbus-binding-tool-glib.c, glib/dbus-gmain.c,
glib/dbus-gsignature.c, glib/dbus-gtype-specialized.c,
glib/dbus-gtype-specialized.h, glib/dbus-gvalue-utils.c,
glib/dbus-gvalue-utils.h, glib/dbus-gvalue.c:
Patch from Rob Taylor <rob.taylor@collabora.co.uk> to add a big
missing piece of the glib bindings jigsaw puzzle. This modifies
the existing specialised types to have N type parameters (rather
than the current 1 or 2 for arrays and dictionaries respectively).
You can then use this to get a glib type to represent any arbitrary
D-Bus struct type using dbus_g_type_get_struct. The only
implementation of these types is with GValueArrays as before,
but it's now possible to store these in arrays, emit them in
signals, etc.
* bus/connection.c (bus_connections_expect_reply): Make
pending reply limit not common to all connections (Bug #5416)
Patch from Kimmo Hämäläinen <kimmo.hamalainen at nokia.com>
* glib/dbus-binding-tool-glib.[ch]: Patch based on Ricardo Kekki's
patch to use an annotation org.freedesktop.DBus.GLib.ClientCSymbol
when generating the client-side methods, instead of overloading
CSymbol which broke everything horribly. My apologies.
* glib/dbus-gtype-specialized.[ch], glib/dbus-gvalue-utils.c: Patch
by me and Rob Taylor to add a simple_free function to D-Bus map
and collection types, which allows those types which can be freed
with a GDestroyNotify (such as GHashTables and GArrays, but not
GPtrArrays) to be stored as the values in hashtables.
* test/glib/test-dbus-glib.c, test/glib/test-service-glib.{c,xml}:
Patch by Rob Taylor to add nested dicts to the glib tests to check
the above code works, and appears not to leak when called repeatedly.
* glib/dbus-binding-tool-glib.c: Patch from Rob Taylor
<rob.taylor@collabora.co.uk> to add support for generating bindings
to arrays that are represented as GPtrArrays rather than GArrays (ie
size-variable things, such as strings, objects, structs, etc).
* dbus/dbus-glib.h, glib/dbus-gproxy.c: Patch from Ricardo Kekki
<ricardo.kekki@movial.fi> to make it possible to inherit from
DBusGProxy, by splitting the DBusGProxy struct into a public part and
a private part, and moving the setting of the DBusGProxyManager into a
connection property, allowing proper GObject construction.
* glib/dbus-binding-tool-glib.c: Patch from Ricardo Kekki
<ricardo.kekki@movial.fi> to make dbus-binding-tool heed C symbol name
annotations when generating glib client bindings.
* qt/qtconnection.cpp (requestName): Changed PROHIBIT_REPLACE to ALLOW_REPLACE
Note - this code is wrong and needs to be fixed by the Qt binding
developers. The flags should be treated as bitfields and not enums.
* qt/qtconnection.h: Change ProhibitReplace to AllowReplace
* dbus/dbus-list.c (link_after): remove #ifdef DBUS_BUILD_TESTS
method for getting messages off the bus in the absence of a
mainloop. This method is much like
dbus_connection_read_write_dispatch except it does not dispatch
the messages to a registered filter function. Instead it
allows a developer to process messages by directly popping
them off the bus.
the parser to continue instead of returning error
(bus_desktop_file_load): Do not free parser data when
parse_section_start or parse_key_value fails because it was
already freed by parser_free (patch from Carlos Garcia Campos
<carlosgc at gnome.org>)
dbus/dbus-server-debug-pipe.c, glib/dbus-binding-tool-glib.c
glib/dbus-glib-tool.c, glib/dbus-gparser.c, glib/dbus-gproxy.c
test/test-segfault.c, test/test-utils.c,
test/glib/test-dbus-glib.c, tools/dbus-cleanup-sockets.c
tools/dbus-launch.c, tools/dbus-tree-view.c, tools/dbus-viewer.c:
Various cleanup of dead code and compiler warnings (patch from
Kjartan Maraas <kmaraas at gnome.org>)
