platform/upstream/dbus.git
19 years ago2005-01-12 Joe Shaw <joeshaw@novell.com>
Joe Shaw [Wed, 12 Jan 2005 22:52:34 +0000 (22:52 +0000)]
2005-01-12  Joe Shaw  <joeshaw@novell.com>

* mono/Makefile.am, mono/example/Makefile.am: Always build the
dbus DLL with --debug.  Clean up after the .mdb files this leaves
behind.

19 years ago2005-01-12 Joe Shaw <joeshaw@novell.com>
Joe Shaw [Wed, 12 Jan 2005 22:48:20 +0000 (22:48 +0000)]
2005-01-12  Joe Shaw  <joeshaw@novell.com>

* mono/Makefile.am, mono/example/Makefile.am: Always build the
dbus DLL with --debug.  Clean up after the .mdb files this leaves
behind.

19 years ago2005-01-12 Joe Shaw <joeshaw@novell.com>
Joe Shaw [Wed, 12 Jan 2005 22:41:29 +0000 (22:41 +0000)]
2005-01-12  Joe Shaw  <joeshaw@novell.com>

* mono/doc/Makefile.am: Need to uninstall the docs on "make
uninstall"

19 years ago2005-01-12 Joe Shaw <joeshaw@novell.com>
Joe Shaw [Wed, 12 Jan 2005 22:34:28 +0000 (22:34 +0000)]
2005-01-12  Joe Shaw  <joeshaw@novell.com>

* mono/Makefile.am, mono/example/Makefile.am: Always build the
dbus DLL with --debug.

19 years ago2005-01-12 Joe Shaw <joeshaw@novell.com>
Joe Shaw [Wed, 12 Jan 2005 17:31:08 +0000 (17:31 +0000)]
2005-01-12  Joe Shaw  <joeshaw@novell.com>

* mono/Makefile.am: Always build the dbus DLL with --debug.

* mono/Arguments.cs (GetDBusTypeConstructor): If the type
is an enum, get the enum's underlying type.  Another mono
1.1.3 fix.

19 years agofix doh! in test for double parsing
Havoc Pennington [Tue, 11 Jan 2005 22:35:58 +0000 (22:35 +0000)]
fix doh! in test for double parsing

19 years ago2005-01-11 Joe Shaw <joeshaw@novell.com>
Joe Shaw [Tue, 11 Jan 2005 22:09:16 +0000 (22:09 +0000)]
2005-01-11  Joe Shaw  <joeshaw@novell.com>

Patch from Sjoerd Simons <sjoerd@luon.net>

* mono/Makefile.am, mono/example/Makefile.am: Don't redefine
DESTDIR.  It breaks stuff.

19 years ago2005-01-11 Joe Shaw <joeshaw@novell.com>
Joe Shaw [Tue, 11 Jan 2005 19:59:06 +0000 (19:59 +0000)]
2005-01-11  Joe Shaw  <joeshaw@novell.com>

Patch from Tambet Ingo <tambet@ximian.com>

* mono/DBusType/Array.cs (Get): Get the underlying element type by
calling type.GetElementType().  The code previously depended on
broken Mono behavior, which was fixed in Mono 1.1.3.

* mono/DBusType/Dict.cs (constructor): Fix the parameters for
Activator.CreateInstance() so that the class's constructor is
called with the right parameters.

19 years ago2005-01-11 Joe Shaw <joeshaw@novell.com>
Joe Shaw [Tue, 11 Jan 2005 19:31:56 +0000 (19:31 +0000)]
2005-01-11  Joe Shaw  <joeshaw@novell.com>

Patch from Timo Teräs <ext-timo.teras@nokia.com>

* dbus/dbus-connection.c
(_dbus_connection_queue_received_message_link): Call
_dbus_connection_remove_timeout() instead of the _locked()
variant, since it's always called from
_dbus_connection_handle_watch(), which handles the locking.
Removed the _locked() variant since it's no longer used.

19 years ago2005-01-03 Havoc Pennington <hp@redhat.com>
Havoc Pennington [Mon, 3 Jan 2005 17:18:11 +0000 (17:18 +0000)]
2005-01-03  Havoc Pennington  <hp@redhat.com>

* dbus/dbus-internals.h: I'm an idiot, _dbus_assert certainly can
return

19 years ago2004-12-26 Havoc Pennington <hp@redhat.com>
Havoc Pennington [Mon, 3 Jan 2005 17:06:35 +0000 (17:06 +0000)]
2004-12-26  Havoc Pennington  <hp@redhat.com>

* dbus/dbus-internals.h: add _DBUS_GNUC_NORETURN to _dbus_assert

19 years agoadd a file to hold the validation routines
Havoc Pennington [Mon, 3 Jan 2005 17:05:26 +0000 (17:05 +0000)]
add a file to hold the validation routines

19 years ago2005-01-03 Havoc Pennington <hp@redhat.com>
Havoc Pennington [Mon, 3 Jan 2005 07:16:12 +0000 (07:16 +0000)]
2005-01-03  Havoc Pennington  <hp@redhat.com>

* dbus/dbus-sysdeps.c (_dbus_sysdeps_test): fix using == on
floating point

* dbus/dbus-string.c (_dbus_string_insert_alignment): new function

19 years agofinish off my TODO list for stuff needed to port dbus-message.c. Next
Havoc Pennington [Mon, 3 Jan 2005 07:15:38 +0000 (07:15 +0000)]
finish off my TODO list for stuff needed to port dbus-message.c. Next
step is to do so.

19 years agojust when the file was looking all beautiful, add horrible hacky code to
Havoc Pennington [Mon, 3 Jan 2005 04:25:32 +0000 (04:25 +0000)]
just when the file was looking all beautiful, add horrible hacky code to
fixup array lengths after setting a value somewhere within the array.

19 years agocleanup args to read func in test suite
Havoc Pennington [Mon, 3 Jan 2005 00:23:03 +0000 (00:23 +0000)]
cleanup args to read func in test suite

19 years agocheckpoint set_variable_length work
Havoc Pennington [Mon, 3 Jan 2005 00:01:59 +0000 (00:01 +0000)]
checkpoint set_variable_length work

19 years agovalues-only DBusTypeWriter
Havoc Pennington [Sun, 2 Jan 2005 22:13:26 +0000 (22:13 +0000)]
values-only DBusTypeWriter

19 years agominor cleanup
Havoc Pennington [Sun, 2 Jan 2005 20:15:01 +0000 (20:15 +0000)]
minor cleanup

19 years agoclean up the naming of dbus_marshal functions
Havoc Pennington [Sun, 2 Jan 2005 19:10:16 +0000 (19:10 +0000)]
clean up the naming of dbus_marshal functions

19 years agodelete some more noise, put args in consistent order (a big bug trap sadly),
Havoc Pennington [Sun, 2 Jan 2005 08:46:42 +0000 (08:46 +0000)]
delete some more noise, put args in consistent order (a big bug trap sadly),
and work on the set value stuff some

19 years agocleanups and fixes
Havoc Pennington [Sun, 2 Jan 2005 05:01:07 +0000 (05:01 +0000)]
cleanups and fixes

19 years agoimplement _dbus_type_writer_write_reader() (to copy a block of values)
Havoc Pennington [Sun, 2 Jan 2005 04:29:52 +0000 (04:29 +0000)]
implement _dbus_type_writer_write_reader() (to copy a block of values)
which is the first step toward a sane reimplementation of all that delete/modify
and realign nonsense in dbus-message.c

19 years agochange strings to be marshaled consistently with the other basic types (address of...
Havoc Pennington [Sun, 2 Jan 2005 02:47:51 +0000 (02:47 +0000)]
change strings to be marshaled consistently with the other basic types (address of the value passed in, so char**)

19 years agomake reader const-correct
Havoc Pennington [Sun, 2 Jan 2005 02:38:23 +0000 (02:38 +0000)]
make reader const-correct

19 years agoextend unit test to display progress and to verify that we are inserting not appendin...
Havoc Pennington [Sun, 2 Jan 2005 02:17:05 +0000 (02:17 +0000)]
extend unit test to display progress and to verify that we are inserting not appending in all cases

19 years agoadd DBusTypeMark
Havoc Pennington [Sun, 2 Jan 2005 00:37:24 +0000 (00:37 +0000)]
add DBusTypeMark

19 years ago2005-01-01 Havoc Pennington <hp@redhat.com>
Havoc Pennington [Sat, 1 Jan 2005 23:32:45 +0000 (23:32 +0000)]
2005-01-01  Havoc Pennington  <hp@redhat.com>

* configure.in: add -Wfloat-equal

19 years agochop dbus-marshal-basic in half and move it to be insertion rather than append based
Havoc Pennington [Sat, 1 Jan 2005 21:05:48 +0000 (21:05 +0000)]
chop dbus-marshal-basic in half and move it to be insertion rather than append based

19 years agovoc Pennington <hp@redhat.com>
Havoc Pennington [Sat, 1 Jan 2005 20:53:42 +0000 (20:53 +0000)]
voc Pennington  <hp@redhat.com>

* dbus/dbus-sysdeps.h: add _DBUS_DOUBLES_BITWISE_EQUAL macro,
for a variety of reasons '==' doesn't do this.

19 years agoall types now unit tested and working.
Havoc Pennington [Sat, 1 Jan 2005 08:19:28 +0000 (08:19 +0000)]
all types now unit tested and working.

19 years agoadd more "subclassing" to the test node classes
Havoc Pennington [Sat, 1 Jan 2005 03:37:26 +0000 (03:37 +0000)]
add more "subclassing" to the test node classes

19 years agoyay, variant values working before the new year.
Havoc Pennington [Sat, 1 Jan 2005 03:28:08 +0000 (03:28 +0000)]
yay, variant values working before the new year.

19 years agominor tweaking
Havoc Pennington [Sat, 1 Jan 2005 02:47:13 +0000 (02:47 +0000)]
minor tweaking

19 years agoadd int64 to test suites, fix minor bug where marshaling assumed that DBusOctets8...
Havoc Pennington [Sat, 1 Jan 2005 02:35:09 +0000 (02:35 +0000)]
add int64 to test suites, fix minor bug where marshaling assumed that DBusOctets8 was aligned when passed in as a function argument

19 years agoremove old unit test code, and clean up whitespace
Havoc Pennington [Sat, 1 Jan 2005 02:03:50 +0000 (02:03 +0000)]
remove old unit test code, and clean up whitespace

19 years agotest framework extension
Havoc Pennington [Sat, 1 Jan 2005 02:02:32 +0000 (02:02 +0000)]
test framework extension

19 years agofurther extend tests
Havoc Pennington [Fri, 31 Dec 2004 22:04:27 +0000 (22:04 +0000)]
further extend tests

19 years agonew test framework in place
Havoc Pennington [Fri, 31 Dec 2004 21:08:06 +0000 (21:08 +0000)]
new test framework in place

19 years ago2004-12-31 Havoc Pennington <hp@redhat.com>
Havoc Pennington [Fri, 31 Dec 2004 21:01:56 +0000 (21:01 +0000)]
2004-12-31  Havoc Pennington  <hp@redhat.com>

* dbus/dbus-string.c (_dbus_string_equal_substrings): new function
I keep wishing I had

19 years agouse vtable instead of flags for types only iterators
Havoc Pennington [Fri, 31 Dec 2004 00:24:24 +0000 (00:24 +0000)]
use vtable instead of flags for types only iterators

19 years agoadd "types only" TypeReader, will move to use a vtable instead of a flag in a minute
Havoc Pennington [Fri, 31 Dec 2004 00:06:34 +0000 (00:06 +0000)]
add "types only" TypeReader, will move to use a vtable instead of a flag in a minute

19 years ago- add variant reader
Havoc Pennington [Thu, 30 Dec 2004 23:34:23 +0000 (23:34 +0000)]
- add variant reader
- further squish the iterator structs

19 years agostore writer container_type in one byte to shrink iterator
Havoc Pennington [Thu, 30 Dec 2004 22:46:13 +0000 (22:46 +0000)]
store writer container_type in one byte to shrink iterator

19 years agoremove unused element_type_len field
Havoc Pennington [Thu, 30 Dec 2004 22:44:44 +0000 (22:44 +0000)]
remove unused element_type_len field

19 years ago- remove DICT type
Havoc Pennington [Thu, 30 Dec 2004 22:34:17 +0000 (22:34 +0000)]
- remove DICT type
- add SIGNATURE type
- fix max name len of 256, should be 255 so it fits in a byte if it's going to be a number
  near 256
- generalize some of the recursive type marshaling code to share between arrays and variants
- just started on implementing variant

19 years ago2004-12-30 John (J5) Palmieri <johnp@redhat.com>
John (J5) Palmieri [Thu, 30 Dec 2004 18:28:03 +0000 (18:28 +0000)]
2004-12-30  John (J5) Palmieri  <johnp@redhat.com>

* python/dbus.py: s/ACTIVATION_REPLY_ACTIVE/ACTIVATION_REPLY_ACTIVATED

19 years ago2004-12-30 John (J5) Palmieri <johnp@redhat.com>
John (J5) Palmieri [Thu, 30 Dec 2004 14:19:54 +0000 (14:19 +0000)]
2004-12-30  John (J5) Palmieri  <johnp@redhat.com>

        * python/dbus_bindings.pyx.in: Change DBUS_ACTIVATION_REPLY_ACTIVATED
        and DBUS_ACTIVATION_REPLY_ALREADY_ACTIVE to match the values in
        dbus-protocol.h.  Because they are defines and not enums they are not
        autogenerated.

19 years agoDBusTypeWriterClass isn't going to work well, so remove vestiges of it.
Havoc Pennington [Wed, 29 Dec 2004 22:31:01 +0000 (22:31 +0000)]
DBusTypeWriterClass isn't going to work well, so remove vestiges of it.

19 years agoswitch to a vtable approach for the reader
Havoc Pennington [Wed, 29 Dec 2004 22:22:41 +0000 (22:22 +0000)]
switch to a vtable approach for the reader

19 years agoput it in the test suite arrays (ok, I need to rewrite the unit tests here to be...
Havoc Pennington [Wed, 29 Dec 2004 06:39:32 +0000 (06:39 +0000)]
put it in the test suite arrays (ok, I need to rewrite the unit tests here to be more automated...)

19 years agostruct of array of struct and array of struct of array
Havoc Pennington [Wed, 29 Dec 2004 06:27:59 +0000 (06:27 +0000)]
struct of array of struct and array of struct of array

19 years agocheckpoint with array-of-struct working
Havoc Pennington [Wed, 29 Dec 2004 05:04:01 +0000 (05:04 +0000)]
checkpoint with array-of-struct working

19 years agoarrays are working, woot
Havoc Pennington [Tue, 28 Dec 2004 23:14:31 +0000 (23:14 +0000)]
arrays are working, woot

19 years agosync busted marshaling work in progress
Havoc Pennington [Tue, 28 Dec 2004 04:55:52 +0000 (04:55 +0000)]
sync busted marshaling work in progress

19 years ago2004-12-26 John (J5) Palmieri <johnp@redhat.com>
John (J5) Palmieri [Mon, 27 Dec 2004 16:01:13 +0000 (16:01 +0000)]
2004-12-26  John (J5) Palmieri  <johnp@redhat.com>

* python/dbus_bindings.pyx.in (bus_activate_service): Bind
dbus_bus_activate_service

* python/dbus.py (Bus.activate_service): activate a service on the
bus.

19 years ago2004-12-24 Havoc Pennington <hp@redhat.com>
Havoc Pennington [Fri, 24 Dec 2004 17:54:01 +0000 (17:54 +0000)]
2004-12-24  Havoc Pennington  <hp@redhat.com>

* test/decode-gcov.c: change to use .gcno and .gcda files, but the
file format has also changed and I haven't adapted to that yet

* Makefile.am: load .gcno files from latest gcc

19 years ago* Forgot to add ChangeLog on last commit
John (J5) Palmieri [Thu, 23 Dec 2004 17:22:29 +0000 (17:22 +0000)]
* Forgot to add ChangeLog on last commit

19 years ago* patch from Rob Taylor <robtaylor@fastmail.fm>
John (J5) Palmieri [Thu, 23 Dec 2004 00:50:37 +0000 (00:50 +0000)]
* patch from Rob Taylor <robtaylor@fastmail.fm>
- wrap bus_get_unix_user method in low level bindings
- add get_unix_user method to the Bus class
- fix extract.py so it can handle unsigned long return types

19 years ago* dbus/make-dbus-glib-error-enum.sh: omit the function keyword for
Olivier Andrieu [Tue, 21 Dec 2004 10:51:46 +0000 (10:51 +0000)]
* dbus/make-dbus-glib-error-enum.sh: omit the function keyword for
  better POSIX compliance.

19 years ago2004-12-19 Havoc Pennington <hp@redhat.com>
Havoc Pennington [Sun, 19 Dec 2004 16:23:09 +0000 (16:23 +0000)]
2004-12-19  Havoc Pennington  <hp@redhat.com>

* dbus/dbus-string.c (_dbus_string_insert_4_aligned)
(_dbus_string_insert_8_aligned): new functions

* dbus/dbus-string.c (_dbus_string_alloc_space): new function

19 years agoback up some files I'm working on that don't do anything yet
Havoc Pennington [Sun, 19 Dec 2004 00:11:49 +0000 (00:11 +0000)]
back up some files I'm working on that don't do anything yet

19 years ago2004-12-18 Havoc Pennington <hp@redhat.com>
Havoc Pennington [Sun, 19 Dec 2004 00:11:20 +0000 (00:11 +0000)]
2004-12-18  Havoc Pennington  <hp@redhat.com>

* dbus/dbus-string.c (_dbus_string_validate_ascii): use ISASCII
macro

* dbus/dbus-message.c: fix a comment, and add a still-unused
not-implemented function

* dbus/dbus-marshal.h: fix comment

* dbus/dbus-internals.h (_DBUS_ISASCII): new macro

19 years ago2004-12-17 Joe Shaw <joeshaw@novell.com>
Joe Shaw [Fri, 17 Dec 2004 22:21:16 +0000 (22:21 +0000)]
2004-12-17  Joe Shaw  <joeshaw@novell.com>

* mono/DBusType/Byte.cs, mono/DBusType/Int32.cs,
mono/DBusType/Int64.cs, mono/DBusType/UInt32.cs,
mono/DBusType/UInt64.cs: Use Enum.GetUnderlyingType() instead of
Type.UnderlyingSystemType to get the actual system type
underneath.  This code previously depended on the broken Mono
behavior, which was fixed in 1.1.3.

19 years ago2004-11-27 Havoc Pennington <hp@redhat.com>
Havoc Pennington [Sat, 27 Nov 2004 07:30:22 +0000 (07:30 +0000)]
2004-11-27  Havoc Pennington  <hp@redhat.com>

* dbus/dbus-string.h (_dbus_string_get_byte): inline when asserts
are disabled
(_dbus_string_get_const_data): inline when asserts are disabled

* dbus/dbus-message.c: record the _dbus_current_generation of
creation so we can complain if dbus_shutdown() is used improperly.
Do this only if checks are enabled.

* dbus/dbus-connection.c: ditto

19 years ago2004-11-26 Havoc Pennington <hp@redhat.com>
Havoc Pennington [Sat, 27 Nov 2004 02:18:36 +0000 (02:18 +0000)]
2004-11-26  Havoc Pennington  <hp@redhat.com>

* test/glib/test-profile.c: add with_bus mode to profile echoes
that go through the bus.

* test/glib/run-test.sh: add ability to run test-profile

* bus/dbus-daemon-1.1.in: fix to say that SIGHUP causes partial
config file reload.

19 years ago2004-11-26 Havoc Pennington <hp@redhat.com>
Havoc Pennington [Sat, 27 Nov 2004 00:19:49 +0000 (00:19 +0000)]
2004-11-26  Havoc Pennington  <hp@redhat.com>

* test/glib/test-profile.c: clean up how the fake_malloc_overhead
thing was implemented

19 years ago2004-11-26 Havoc Pennington <hp@redhat.com>
Havoc Pennington [Fri, 26 Nov 2004 07:42:24 +0000 (07:42 +0000)]
2004-11-26  Havoc Pennington  <hp@redhat.com>

* test/glib/test-profile.c: tweak a bit, add support for some
made-up minimal malloc overhead with plain sockets, since in
real life some sort of buffers are unavoidable thus we could
count them in the theoretical best case

19 years ago2004-11-26 Havoc Pennington <hp@redhat.com>
Havoc Pennington [Fri, 26 Nov 2004 07:03:05 +0000 (07:03 +0000)]
2004-11-26  Havoc Pennington  <hp@redhat.com>

* dbus/dbus-message.c (dbus_message_cache_or_finalize): fix bug
where I was trying to cache one too many messages

19 years ago2004-11-26 Havoc Pennington <hp@redhat.com>
Havoc Pennington [Fri, 26 Nov 2004 06:52:15 +0000 (06:52 +0000)]
2004-11-26  Havoc Pennington  <hp@redhat.com>

* dbus/dbus-message.c: reimplement message cache as an array which
makes the cache about twice as fast and saves maybe 1.5% overall

19 years ago2004-11-26 Havoc Pennington <hp@redhat.com>
Havoc Pennington [Fri, 26 Nov 2004 06:30:38 +0000 (06:30 +0000)]
2004-11-26  Havoc Pennington  <hp@redhat.com>

* dbus/dbus-threads.c (init_global_locks): forgot to put the
message cache lock here

19 years ago2004-11-26 Havoc Pennington <hp@redhat.com>
Havoc Pennington [Fri, 26 Nov 2004 06:22:53 +0000 (06:22 +0000)]
2004-11-26  Havoc Pennington  <hp@redhat.com>

* dbus/dbus-message.c (struct DBusMessage): put the locked bit and
the "char byte_order" next to each other to save 4 bytes
(dbus_message_new_empty_header): reduce preallocation, since the
message cache should achieve a similar effect
(dbus_message_cache_or_finalize, dbus_message_get_cached): add a
message cache that keeps a few DBusMessage around in a pool,
another 8% speedup or so.

* dbus/dbus-dataslot.c (_dbus_data_slot_list_clear): new function

19 years ago2004-11-25 Havoc Pennington <hp@redhat.com>
Havoc Pennington [Fri, 26 Nov 2004 02:29:00 +0000 (02:29 +0000)]
2004-11-25  Havoc Pennington  <hp@redhat.com>

* dbus/dbus-transport-unix.c (unix_do_iteration): if we're going
to write, without reading or blocking, try it before the poll()
and skip the poll() if nothing remains to write. This is about a
3% speedup in the echo client/server

19 years ago2004-11-25 Havoc Pennington <hp@redhat.com>
Havoc Pennington [Fri, 26 Nov 2004 01:53:13 +0000 (01:53 +0000)]
2004-11-25  Havoc Pennington  <hp@redhat.com>

        The primary change here is to always write() once before adding
the write watch, which gives us about a 10% performance increase.

* dbus/dbus-transport-unix.c: a number of modifications to cope
with removing messages_pending
(check_write_watch): properly handle
DBUS_AUTH_STATE_WAITING_FOR_MEMORY; adapt to removal of
messages_pending stuff
(check_read_watch): properly handle WAITING_FOR_MEMORY and
AUTHENTICATED cases
(unix_handle_watch): after writing, see if the write watch can be
removed
(unix_do_iteration): assert that write_watch/read_watch are
non-NULL rather than testing that they aren't, since they
aren't allowed to be NULL. check_write_watch() at the end so
we add the watch if we did not finish writing (e.g. got EAGAIN)

* dbus/dbus-transport-protected.h: remove messages_pending call,
since it resulted in too much inefficient watch adding/removing;
instead we now require that the transport user does an iteration
after queueing outgoing messages, and after trying the first
write() we add a write watch if we got EAGAIN or exceeded our
max bytes to write per iteration setting

* dbus/dbus-string.c (_dbus_string_validate_signature): add this
function

* dbus/dbus-server-unix.c (unix_finalize): the socket name was
freed and then accessed, valgrind flagged this bug, fix it

* dbus/dbus-message.c: fix several bugs where HEADER_FIELD_LAST was taken
as the last valid field plus 1, where really it is equal to the
last valid field. Corrects some message corruption issues.

* dbus/dbus-mainloop.c: verbosity changes

* dbus/dbus-keyring.c (_dbus_keyring_new_homedir): handle OOM
instead of aborting in one of the test codepaths

* dbus/dbus-internals.c (_dbus_verbose_real): fix a bug that
caused not printing the pid ever again if a verbose was missing
the newline at the end
(_dbus_header_field_to_string): add HEADER_FIELD_SIGNATURE

* dbus/dbus-connection.c: verbosity changes;
(dbus_connection_has_messages_to_send): new function
(_dbus_connection_message_sent): no longer call transport->messages_pending
(_dbus_connection_send_preallocated_unlocked): do one iteration to
try to write() immediately, so we can avoid the write watch. This
is the core purpose of this patchset
(_dbus_connection_get_dispatch_status_unlocked): if disconnected,
dump the outgoing message queue, so nobody will get confused
trying to send them or thinking stuff is pending to be sent

* bus/test.c: verbosity changes

* bus/driver.c: verbosity/assertion changes

* bus/dispatch.c: a bunch of little tweaks to get it working again
because this patchset changes when/where you need to block.

19 years ago2004-11-23 Havoc Pennington <hp@redhat.com>
Havoc Pennington [Tue, 23 Nov 2004 06:21:12 +0000 (06:21 +0000)]
2004-11-23  Havoc Pennington  <hp@redhat.com>

* test/glib/test-profile.c: modify to accept a plain_sockets
argument in which case it will bench plain sockets instead of
libdbus, for comparison purposes.

19 years ago2004-11-22 Havoc Pennington <hp@redhat.com>
Havoc Pennington [Mon, 22 Nov 2004 19:20:29 +0000 (19:20 +0000)]
2004-11-22  Havoc Pennington  <hp@redhat.com>

* test/glib/test-profile.c (N_CLIENT_THREADS): run multiple
threads for more time, so sysprof can get a grip on it.

* dbus/dbus-string.c (_dbus_string_validate_utf8): remove
pointless variable

19 years ago2004-11-13 Havoc Pennington <hp@redhat.com>
Havoc Pennington [Sat, 13 Nov 2004 07:07:47 +0000 (07:07 +0000)]
2004-11-13  Havoc Pennington  <hp@redhat.com>

* test/glib/test-profile.c: fix this thing up a bit

* dbus/dbus-message.c (dbus_message_new_empty_header): increase
preallocation sizes by a fair bit; not sure if this will be an
overall performance win or not, but it does reduce reallocs.

* dbus/dbus-string.c (set_length, reallocate_for_length): ignore
the test hack that forced constant realloc if asserts are
disabled, so we can profile sanely. Sprinkle in some
_DBUS_UNLIKELY() which are probably pointless, but before I
noticed the real performance problem I put them in.
(_dbus_string_validate_utf8): micro-optimize this thing a little
bit, though callgrind says it didn't help; then special-case
ascii, which did help a lot; then be sure we detect nul bytes as
invalid, which is a bugfix.
(align_length_then_lengthen): add some more _DBUS_UNLIKELY
superstition; use memset to nul the padding instead of a manual
loop.
(_dbus_string_get_length): inline this as a
macro; it showed up in the profile because it's used for loop
tests and so forth

20 years agoadd a little comment about WNOHANG/EINTR
Havoc Pennington [Wed, 10 Nov 2004 20:12:15 +0000 (20:12 +0000)]
add a little comment about WNOHANG/EINTR

20 years ago2004-11-10 Colin Walters <walters@verbum.org>
Colin Walters [Wed, 10 Nov 2004 06:53:10 +0000 (06:53 +0000)]
2004-11-10  Colin Walters  <walters@verbum.org>

* dbus/dbus-spawn.c (check_babysit_events): Handle EINTR,
for extra paranoia.

20 years ago2004-11-09 Colin Walters <walters@verbum.org>
Colin Walters [Tue, 9 Nov 2004 06:11:33 +0000 (06:11 +0000)]
2004-11-09  Colin Walters  <walters@verbum.org>

* dbus/dbus-string.c (_dbus_string_get_length): New
function, writes DBusString to C buffer.

* dbus/dbus-string.h: Prototype it.

* dbus/dbus-message.c (dbus_message_type_to_string): New
function, converts message type into C string.

* dbus/dbus-message.h: Prototype it.

* bus/selinux.c (bus_selinux_check): Take source pid,
target pid, and audit data.  Pass audit data to
avc_has_perm.
(log_audit_callback): New function, appends extra
audit information.
(bus_selinux_allows_acquire_service): Also take
service name, add it to audit data.
(bus_selinux_allows_send): Also take message
type, interface, method member, error name,
and destination, and add them to audit data.
(log_cb): Initialize func_audit.

* bus/selinux.h (bus_selinux_allows_acquire_service)
(bus_selinux_allows_send): Update prototypes

* bus/services.c (bus_registry_acquire_service): Pass
service name to bus_selinux_allows_acquire_service.

* bus/bus.c (bus_context_check_security_policy): Pass
additional audit data.  Move assignment of dest
to its own line.

20 years ago2004-11-07 Colin Walters <walters@verbum.org>
Colin Walters [Sun, 7 Nov 2004 17:06:34 +0000 (17:06 +0000)]
2004-11-07  Colin Walters  <walters@verbum.org>

* dbus/dbus-transport-unix.c (do_authentication): Always
initialize auth_completed.

20 years ago2004-11-07 Colin Walters <walters@verbum.org>
Colin Walters [Sun, 7 Nov 2004 17:05:19 +0000 (17:05 +0000)]
2004-11-07  Colin Walters  <walters@verbum.org>

* bus/bus.c (load_config): Break into three
separate functions: process_config_first_time_only,
process_config_every_time, and process_config_postinit.
(process_config_every_time): Move call of
bus_registry_set_service_context_table into
process_config_postinit.
(process_config_postinit): New function, does
any processing that needs to happen late
in initialization (and also on reload).
(bus_context_new): Instead of calling load_config,
open config parser here and call process_config_first_time_only
and process_config_every_time directly.  Later, after
we have forked but before changing UID,
invoke bus_selinux_full_init, and then call
process_config_postinit.
(bus_context_reload_config): As in bus_context_new,
load parse file inside here, and call process_config_every_time
and process_config_postinit.

* bus/services.h, bus/services.c
(bus_registry_set_service_context_table): Rename
from bus_registry_set_sid_table.  Take string hash from config
parser, and convert them here into SIDs.

* bus/config-parser.c (struct BusConfigParser): Have
config parser only store a mapping of service->context
string.
(merge_service_context_hash): New function.
(merge_included): Merge context string hashes instead
of using bus_selinux_id_table_union.
(bus_config_parser_new): Don't use bus_selinux_id_table_new;
simply create a new string hash.
(bus_config_parser_unref): Unref it.
(start_selinux_child): Simply insert strings into hash,
don't call bus_selinux_id_table_copy_over.

* bus/selinux.h, bus/selinux.c (bus_selinux_id_table_union)
(bus_selinux_id_table_copy_over): Delete.

20 years ago*** empty log message ***
Colin Walters [Thu, 4 Nov 2004 04:10:18 +0000 (04:10 +0000)]
*** empty log message ***

20 years ago2004-11-03 Colin Walters <walters@verbum.org>
Colin Walters [Thu, 4 Nov 2004 04:09:31 +0000 (04:09 +0000)]
2004-11-03  Colin Walters  <walters@verbum.org>

* bus/selinux.c (bus_selinux_pre_init): Kill some unused
variables.

20 years ago2004-11-03 Colin Walters <walters@verbum.org>
Colin Walters [Wed, 3 Nov 2004 16:38:35 +0000 (16:38 +0000)]
2004-11-03  Colin Walters  <walters@verbum.org>

* bus/test-main.c (test_pre_hook): Fix test logic,
thanks Joerg Barfurth <Joerg.Barfurth@Sun.COM>.

20 years ago*** empty log message ***
Colin Walters [Tue, 2 Nov 2004 20:28:00 +0000 (20:28 +0000)]
*** empty log message ***

20 years ago2004-11-02 Colin Walters <walters@verbum.org>
Colin Walters [Tue, 2 Nov 2004 20:27:48 +0000 (20:27 +0000)]
2004-11-02  Colin Walters  <walters@verbum.org>

* bus/selinux.c (bus_selinux_init): Split into two functions,
bus_selinux_pre_init and bus_selinux_post_init.
(bus_selinux_pre_init): Just determine whether SELinux is
enabled.
(bus_selinux_post_init): Do everything else.

* bus/main.c (main): Call bus_selinux_pre_init before parsing
config file, and bus_selinux_post_init after.  This ensures that
we don't lose the policyreload notification thread that
bus_selinux_init created before forking previously.

* bus/test-main.c (test_pre_hook): Update for split.

20 years agoPatch from Johan Fischer to make mono doc install create the doc dir.
Owen Fraser-Green [Tue, 2 Nov 2004 10:19:20 +0000 (10:19 +0000)]
Patch from Johan Fischer to make mono doc install create the doc dir.

20 years ago2004-10-29 Colin Walters <walters@redhat.com>
Colin Walters [Fri, 29 Oct 2004 19:59:15 +0000 (19:59 +0000)]
2004-10-29  Colin Walters  <walters@redhat.com>

* dbus/dbus-sysdeps.h (_dbus_become_daemon): Also take
parameter for fd to write pid to.

* dbus/dbus-sysdeps.c (_dbus_become_daemon): Implement it.

* bus/bus.c (bus_context_new): Pass print_pid_fd
to _dbus_become_daemon (bug #1720)

20 years ago2004-10-29 Colin Walters <walters@redhat.com>
Colin Walters [Fri, 29 Oct 2004 19:22:59 +0000 (19:22 +0000)]
2004-10-29  Colin Walters  <walters@redhat.com>

Patch from Ed Catmur <ed@catmur.co.uk>

* mono/doc/Makefile.am (install-data-local): Handle
DESTDIR.

20 years ago2004-10-29 Colin Walters <walters@redhat.com>
Colin Walters [Fri, 29 Oct 2004 19:20:02 +0000 (19:20 +0000)]
2004-10-29  Colin Walters  <walters@redhat.com>

* bus/.cvsignore, qt/.cvsignore: Update.

20 years ago2004-10-29 Colin Walters <walters@redhat.com>
Colin Walters [Fri, 29 Oct 2004 19:17:57 +0000 (19:17 +0000)]
2004-10-29  Colin Walters  <walters@redhat.com>

Patch from Kristof Vansant <de_lupus@pandora.be>

* configure.in: Detect Slackware.
* bus/Makefile.am (SCRIPT_IN_FILES): Add rc.messagebus.in.
* bus/rc.messagebus.in: New file.

20 years agoAdd comment about workaround.
Colin Walters [Fri, 29 Oct 2004 18:52:30 +0000 (18:52 +0000)]
Add comment about workaround.

20 years ago2004-10-29 Colin Walters <walters@redhat.com>
Colin Walters [Fri, 29 Oct 2004 18:50:29 +0000 (18:50 +0000)]
2004-10-29  Colin Walters  <walters@redhat.com>

* tools/dbus-monitor.c (filter_func): Return
DBUS_HANDLER_RESULT_HANDLED in filter function
for now.  See:
http://freedesktop.org/pipermail/dbus/2004-August/001433.html

20 years ago2004-10-29 Colin Walters <walters@redhat.com>
Colin Walters [Fri, 29 Oct 2004 18:13:53 +0000 (18:13 +0000)]
2004-10-29  Colin Walters  <walters@redhat.com>

Patch from Matthew Rickard <mjricka@epoch.ncsc.mil>

* bus/services.c (bus_registry_acquire_service):
Correctly retrieve service name from DBusString
for printing.

20 years ago2004-10-29 Colin Walters <walters@redhat.com>
Colin Walters [Fri, 29 Oct 2004 17:37:24 +0000 (17:37 +0000)]
2004-10-29 Colin Walters  <walters@redhat.com>

* dbus/dbus-glib.h: Update documentation to not
refer to internal APIs.

20 years ago2004-10-27 Joe Shaw <joeshaw@novell.com>
Joe Shaw [Wed, 27 Oct 2004 22:35:03 +0000 (22:35 +0000)]
2004-10-27  Joe Shaw  <joeshaw@novell.com>

* mono/Arguments.cs (GetDBusTypeConstructor):
type.UnderlyingSystemType will return "System.Byte" if you do it
on "byte[]", which is not what we want.  So check the type.IsArray
property and use System.Array instead.

20 years agoJohn (J5) Palmieri <johnp@redhat.com>
John (J5) Palmieri [Mon, 25 Oct 2004 18:48:58 +0000 (18:48 +0000)]
John (J5) Palmieri  <johnp@redhat.com>

        * dbus/dbus-sysdeps.c (fill_user_info): On errors do not free
        the DBusUserInfo structure since this is passed into the function.
        This would cause a double free when the function that allocated
        the structure would try to free it when an error occured.

        * (bus/session.conf.in, bus/Makefile.am, dbus/configure.in):
        use /usr/share/dbus-1/services instead of /usr/lib/dbus-1.0/services
        for service activation to avoid 32bit/64bit parallel install issues

20 years ago2004-10-21 Colin Walters <walters@verbum.org>
Colin Walters [Fri, 22 Oct 2004 02:14:00 +0000 (02:14 +0000)]
2004-10-21  Colin Walters  <walters@verbum.org>

* AUTHORS: Fix my email address, the @gnu.org one
has been bouncing for some time.  Also add J5.

20 years ago2004-10-21 Colin Walters <walters@verbum.org>
Colin Walters [Fri, 22 Oct 2004 02:07:29 +0000 (02:07 +0000)]
2004-10-21  Colin Walters  <walters@verbum.org>

* dbus/dbus-transport-unix.c (do_authentication): Return
authentication status to callers.
(unix_handle_watch): If we completed authentication this round,
don't do another read.  Instead wait until the next iteration,
after we've read any pending data in the auth buffer.
(unix_do_iteration): Ditto.
(unix_handle_watch): Updated for new do_authentication prototype.