Commit graph

1974 commits

Author SHA1 Message Date
Havoc Pennington
be672187a5 oops, committed some debug junk 2005-02-13 20:23:30 +00:00
Havoc Pennington
7f2973c8d1 2005-02-13 Havoc Pennington <hp@redhat.com>
* dbus/dbus-connection.c (dbus_connection_return_message)
	(dbus_connection_borrow_message): hold dispatch lock while message
	is outstanding
	(_dbus_connection_block_for_reply): hold dispatch lock while we
	block for the reply, so nobody steals our reply
	(dbus_connection_pop_message): hold the dispatch lock while we
	pluck the message
2005-02-13 20:21:59 +00:00
Havoc Pennington
770dfe0571 2005-02-13 Havoc Pennington <hp@redhat.com>
* dbus/dbus-connection.c (_dbus_connection_acquire_dispatch)
	(_dbus_connection_release_dispatch)
	(_dbus_connection_acquire_io_path)
	(_dbus_connection_release_io_path): make the mutex and condvar
	control access to the "acquired" flag. Drop the connection lock
	while waiting on the condvar. Hopefully these are baby steps in
	roughly the right direction.
2005-02-13 19:49:22 +00:00
Havoc Pennington
46ee60868d 2005-02-13 Havoc Pennington <hp@redhat.com>
* dbus/dbus-connection.c: use separate mutexes for the condition
	variables; this is some kind of baseline for sanity, but the
	condition variables still aren't used correctly afaict
2005-02-13 18:17:35 +00:00
Havoc Pennington
f349e6b8c5 2005-02-13 Havoc Pennington <hp@redhat.com>
* dbus/dbus-object-tree.c (handle_default_introspect_and_unlock):
	fix a double-unlock

	* dbus/dbus-connection.c
	(_dbus_connection_detach_pending_call_unlocked): add this

        Initial semi-correct pass through to fix thread locking; there are
	still some issues with the condition variable paths I'm pretty
	sure

	* dbus/dbus-server.c: add a mutex on DBusServer and appropriate
	lock/unlock calls

	* dbus/dbus-connection.c (_dbus_connection_do_iteration_unlocked):
	rename to add _unlocked
	(struct DBusConnection): move "dispatch_acquired" and
	"io_path_acquired" to use only one bit each.
	(CONNECTION_LOCK, CONNECTION_UNLOCK): add checks with !DBUS_DISABLE_CHECKS
	(dbus_connection_set_watch_functions): hacky fix to reentrancy
	(_dbus_connection_add_watch, _dbus_connection_remove_watch)
	(_dbus_connection_toggle_watch, _dbus_connection_add_timeout)
	(_dbus_connection_remove_timeout)
	(_dbus_connection_toggle_timeout): drop lock when calling out to
	user functions; done in a hacky/bad way.
	(_dbus_connection_send_and_unlock): add a missing unlock
	(_dbus_connection_block_for_reply): add a missing unlock

	* dbus/dbus-transport.c (_dbus_transport_get_is_authenticated):
	drop lock in a hacky probably unsafe way to call out to user
	function
2005-02-13 17:16:25 +00:00
Havoc Pennington
970be5fda3 2005-02-12 Havoc Pennington <hp@redhat.com>
* tools/dbus-tree-view.c (info_set_func_text): display more
	details on args

	* bus/driver.c (bus_driver_handle_list_services): list the bus
	driver

	* glib/dbus-gparser.c (parse_arg): generate an arg name if none is supplied

	* glib/dbus-gidl.c (signal_info_get_n_args): new function
	(method_info_get_n_args): new function
2005-02-12 20:27:45 +00:00
Havoc Pennington
85fb9ff93c 2005-02-12 Havoc Pennington <hp@redhat.com>
* bus/driver.c (bus_driver_handle_introspect): add introspection
	for bus driver
2005-02-12 20:13:08 +00:00
Havoc Pennington
5b26893cdd 2005-02-12 Havoc Pennington <hp@redhat.com>
* bus/driver.c: put the signature of each bus driver method in the
	table of handlers and check it on incoming calls; this isn't
	really useful, but going to add introspect support in a minute.
2005-02-12 17:13:13 +00:00
Havoc Pennington
03a4a5a976 some todo and spec tweaks 2005-02-12 04:30:34 +00:00
Havoc Pennington
892c68a727 forgot to commit removing address item from TODO 2005-02-12 03:56:22 +00:00
Joe Shaw
0242b82152 2005-02-11 Joe Shaw <joeshaw@novell.com>
* mono/Connection.cs: The unpredictability of finalizers in mono
	prevents us from deterministically disconnecting the filters from
	the Service class's finalizer, so move tracking of filters and
	matches here.  Add API for that.

	* mono/Service.cs: Remove the code, add code which calls the
	methods now on the Connection class.
2005-02-12 00:51:33 +00:00
John (J5) Palmieri
38dad64ace * python/dbus.py (class Sender): added to support dbus signals better
(Bus::add_signal_receiver): added expand_args parameter which defaults
to True.  When expand args is True the signal handler will pass the
message arguments as parameters to the signal handler.  If False
revert to previous behavior where the signal handler must get the
argument list from the message.  This is to help port applications
like HAL that have a tendancy to send variable length argument lists.
self._match_rule_to_receivers is now a dict of dicts.
(Bus::remove_signal_receiver): pop handler off the dict intead of
removing it from a list
(Bus::_signal_func): change signal handlers so that interface,
signal_name, service, path and message are packed into a Sender
object and that is passed to the handler.  If expand_args is True
extract the args list from the message and append it to the parameter
list

* python/dbus_bindings.pyx.in (class Signature): added to support
signiature types
(MessageIter::__init__): changed iteration limit to match D-BUS
(MessageIter::get*): added INT16, UINT16, SIGNATURE, DICT_ENTRY,
STRUCT and VARIENT type support
(MessageIter::python_value_to_dbus_sig): made recursive to support
recursive types
(MessageIter::append*): added Signature, dict, tuple
support

* python/examples/example-client.py: added examples of getting tuples
and dicts

* python/examples/example-service.py: added examples of sending tuples
and dicts

* python/examples/example-signal-recipient.py: Fixed to handle new
signal callback format
2005-02-11 19:51:18 +00:00
Havoc Pennington
aa4f823781 2005-02-10 Havoc Pennington <hp@redhat.com>
* test/glib/test-dbus-glib.c (main): fix so this test doesn't fail
	(call dbus_g_proxy_add_signal)

	* dbus/dbus-server-unix.c (_dbus_server_new_for_tcp_socket):
	escape the hostname
	(_dbus_server_new_for_domain_socket): escape the path

	* dbus/dbus-address.c (dbus_address_escape_value): new
	(dbus_address_unescape_value): new
	(dbus_parse_address): unescape values

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

	* doc/dbus-specification.xml: explain how to escape values in
	addresses
2005-02-11 03:37:03 +00:00
Havoc Pennington
71f3b461b3 2005-02-10 Havoc Pennington <hp@redhat.com>
* dbus/dbus-message-factory.c (generate_special): modify test to
	avoid using a non-basic dict key

	* dbus/dbus-marshal-validate-util.c: add test for the below

	* doc/dbus-specification.xml: require that dict keys are a basic
	type

	* dbus/dbus-marshal-validate.c
	(_dbus_validate_signature_with_reason): require that dict key is a
	basic type
2005-02-11 01:13:45 +00:00
Havoc Pennington
df0c640880 2005-02-10 Havoc Pennington <hp@redhat.com>
* dbus/dbus-object-tree.c (handle_default_introspect_and_unlock):
	change to be _and_unlock instead of _unlocked

	* dbus/dbus-connection.c
	(_dbus_connection_send_preallocated_unlocked_no_update): rename to
	have no_update so we can find this bug quickly in future
2005-02-10 23:47:55 +00:00
Havoc Pennington
83dd8129a0 2005-02-10 Havoc Pennington <hp@redhat.com>
* dbus/dbus-message-util.c (verify_test_message): tests for string
	array

	* dbus/dbus-message.c (dbus_message_append_args_valist): add
	support for arrays of string/signature/path
2005-02-10 23:17:27 +00:00
Joe Shaw
fd146b370d 2005-02-10 Joe Shaw <joeshaw@novell.com>
* dbus/dbus-connection.c
	(_dbus_connection_queue_received_message_link,
	_dbus_connection_message_sent): Add the path to
	the verbose output.
	(_dbus_connection_send_preallocated_and_unlock): Added.  Calls
	_dbus_connection_send_preallocated_unlocked(), updated the
	dispatch status, and unlocks.  Fixes a bug where certain
	situations (like a broken pipe) could cause a Disconnect message
	to not be sent, tricking the bus into thinking a service was still
	there when the process had quit.
	(_dbus_connection_send_preallocated): Call
	_dbus_connection_send_preallocated_and_unlock().
	(_dbus_connection_send_and_unlock): Added.  Calls
	_dbus_connection_send_preallocated_and_unlock().
	(dbus_connection_send): Call _dbus_connection_send_and_unlock().
	(dbus_connection_send_with_reply): Update the dispatch status and
	unlock.

	* mono/Service.cs (~Service): Added.  Removes the filter so that
	we don't get unmanaged code calling back into a GCed delegate.
	(RemoveFilter); Added.
2005-02-10 23:01:28 +00:00
John (J5) Palmieri
8fda602e99 * dbus/dbus-message.c (dbus_message_iter_open_container):
- Removed check for iterator type being an array because
	get_arg_type does not work with writer iterators
	- Pass NULL to _dbus_type_writer_recurse if signiture is NULL
2005-02-10 04:36:23 +00:00
Colin Walters
01f18bcc67 2005-02-07 Colin Walters <walters@verbum.org>
* bus/selinux.c (bus_selinux_allows_send): Handle NULL for
	sender or proposed_recipient.
2005-02-07 14:14:50 +00:00
Havoc Pennington
bf75b24d26 2005-02-07 Havoc Pennington <hp@redhat.com>
* doc/dbus-specification.xml: some more language cleanups; add
	stuff about how to deal with invalid protocol and extension
	points; add _ to allowed chars in auth commands; add EXTENSION_
	auth command prefix
2005-02-07 05:44:57 +00:00
Colin Walters
b4fbcb98f1 2005-02-07 Colin Walters <walters@verbum.org>
* bus/selinux.c (bus_selinux_allows_send): Handle NULL for
	sender or proposed_recipient.
2005-02-07 05:31:59 +00:00
Havoc Pennington
d76937e2e1 2005-02-06 Havoc Pennington <hp@redhat.com>
* s/expected/required/ in a couple places for clarity
2005-02-07 04:47:52 +00:00
Havoc Pennington
7fd58272f9 fix the docs on signal you get when you lose a bus name 2005-02-06 15:35:25 +00:00
Havoc Pennington
96a4751632 2005-02-06 Havoc Pennington <hp@redhat.com>
* dbus/dbus-message-factory.c (generate_special): more tests

	* dbus/dbus-marshal-validate.c (validate_body_helper): detect
	array length that exceeds the maximum
2005-02-06 05:19:50 +00:00
Havoc Pennington
8f04e1e01f 2005-02-05 Havoc Pennington <hp@redhat.com>
* dbus/dbus-message-factory.c (generate_special): more test cases,
	increasing coverage

	* dbus/dbus-marshal-validate.c (validate_body_helper): return the
	reason why a signature was invalid

	* dbus/dbus-marshal-header.c (load_and_validate_field): fix to
	skip the length of the string before we look at it in validation

	* dbus/dbus-string-util.c (_dbus_string_test): add tests for
	equal_substring

	* dbus/dbus-message.c (_dbus_message_loader_new): default
	max_message_length to DBUS_MAXIMUM_MESSAGE_LENGTH
2005-02-06 04:21:57 +00:00
Havoc Pennington
0e53d4eed3 2005-02-05 Havoc Pennington <hp@redhat.com>
* dbus/dbus-marshal-validate.c (validate_body_helper): fix crash
	if the signature of a variant was empty
	(_dbus_validate_signature_with_reason): catch "(a)" (array inside
	struct with no element type)

	* dbus/dbus-message-factory.c (generate_uint32_changed): add more
	mangled messages to break things
2005-02-05 21:03:28 +00:00
Havoc Pennington
dde6f303e5 2005-02-04 Havoc Pennington <hp@redhat.com>
* glib/dbus-gproxy.c (dbus_g_proxy_disconnect_signal): use
	g_quark_try_string() so it actually can return 0
	(dbus_g_proxy_connect_signal): ditto

2005-02-04  Havoc Pennington  <hp@redhat.com>

	* glib/dbus-gproxy.c (dbus_g_proxy_emit_remote_signal): fix a
	bogus warning
	(tristring_from_message): assert cleanly on null path/interface
	(should not be possible though I decided later)
	(dbus_g_proxy_dispose): move proxy manager unregistration here
	(DBUS_G_PROXY_DESTROYED): add this macro, and use it in a bunch of
	g_return_if_fail() checks
2005-02-05 04:15:57 +00:00
Havoc Pennington
849493eafb 2005-02-04 Havoc Pennington <hp@redhat.com>
* doc/Makefile.am (EXTRA_DIST): add DTDs to makefile

	* doc/introspect.dtd: add introspect.dtd from David A. Wheeler
	(with some minor changes)

	* doc/dbus-specification.xml: add deprecated attribute to
	introspection format
2005-02-05 03:24:54 +00:00
Havoc Pennington
671246e6ac 2005-01-31 Havoc Pennington <hp@redhat.com>
* glib/dbus-gproxy.c: rewrite how signals work again, this time I
	think it's sort of right
2005-01-31 23:17:19 +00:00
Havoc Pennington
4f0fd00e9e 2005-01-30 Havoc Pennington <hp@redhat.com>
* tools/dbus-viewer.c: kind of half-ass hook up the option menu.
2005-01-31 03:15:24 +00:00
Havoc Pennington
d2c1a633d1 2005-01-30 Havoc Pennington <hp@redhat.com>
* tools/dbus-names-model.c: dynamically watch NameOwnerChanged

	* autogen.sh: change to autotools 1.9

	* glib/dbus-gproxy.c: completely change how signals work
	(dbus_g_proxy_add_signal): new function to specify signature of a
	signal
	(dbus_g_proxy_emit_received): marshal the dbus message to GValues,
	and g_warning if the incoming message has the wrong signature.
2005-01-31 02:55:12 +00:00
Havoc Pennington
d5b7d7a78c 2005-01-30 Havoc Pennington <hp@redhat.com>
* tools/dbus-names-model.c (have_names_notify): fix this

	* dbus/dbus-message.c (_dbus_message_iter_get_args_valist): clean
	up the string array handling a bit
2005-01-30 23:29:50 +00:00
Havoc Pennington
1dcacffc32 2005-01-30 Havoc Pennington <hp@redhat.com>
* glib/dbus-glib.c (dbus_g_pending_call_set_notify): new function
	(dbus_g_pending_call_cancel): new function

	* dbus/dbus-glib.h: move GType decls for connection/message here;
	* dbus/dbus-glib.c: move all the g_type and ref/unref stuff in
	here, just kind of rationalizing how we handle all that

	* tools/dbus-names-model.c: new file for a tree model listing the
	services on a bus

	* tools/dbus-tree-view.c (model_new): use proper typing on the
	model rows
2005-01-30 23:06:32 +00:00
Havoc Pennington
4d985d9890 2005-01-30 Havoc Pennington <hp@redhat.com>
* glib/dbus-gmain.c: add a custom GSource back that just checks
	whether the message queue has anything in it; otherwise, there are
	cases where we won't see messages in the queue since there was no
	IO visible to the glib main loop

	* dbus/dbus-connection-internal.h (_DBUS_DEFAULT_TIMEOUT_VALUE):
	increase default message timeout to 25 seconds
2005-01-30 20:06:52 +00:00
Havoc Pennington
191adf3530 2005-01-30 Havoc Pennington <hp@redhat.com>
* test/glib/test-profile.c (no_bus_stop_server): remove the
	warning about the g_warning that I just fixed
2005-01-30 19:35:00 +00:00
Havoc Pennington
96ba08f456 2005-01-30 Havoc Pennington <hp@redhat.com>
* glib/dbus-gmain.c: rewrite the main loop stuff to avoid the
	custom source, seems to be a lot easier to understand and work
	better.
2005-01-30 19:33:29 +00:00
Havoc Pennington
6b8ffc69ad 2005-01-30 Havoc Pennington <hp@redhat.com>
I think this main loop thing is conceptually broken, but here are
	some band aids. I'll maybe rewrite it in a minute.

	* glib/dbus-gmain.c (add_timeout): timeout stuff doesn't use the
	custom GSource, so don't pass it in; confusing
	(gsource_server_finalize, gsource_connection_finalize): add
	finalize handlers that remove all the watches.
2005-01-30 18:25:14 +00:00
Havoc Pennington
98ad8a8ec6 2005-01-30 Havoc Pennington <hp@redhat.com>
* glib/dbus-gobject.c (introspect_properties): fix the XML
	generated

	* dbus/dbus-message.c (dbus_message_unref): add an in_cache flag
	which effectively detects the use of freed messages

	* glib/dbus-gobject.c (handle_introspect): modify and return the
	reply message instead of the incoming message

	* dbus/dbus-object-tree.c (handle_default_introspect_unlocked):
	gee, maybe it should SEND THE XML instead of just making a string
	and freeing it again ;-)

	* tools/dbus-print-message.c (print_message): improve printing of
	messages

	* configure.in: add debug-glib.service to the output
2005-01-30 07:44:08 +00:00
Havoc Pennington
41f52c96d6 2005-01-30 Havoc Pennington <hp@redhat.com>
dbus-viewer introspected and displayed the bus driver

	* dbus/dbus-object-tree.c
	(object_tree_test_iteration): add tests for a handler registered on "/"

	* dbus/dbus-object-tree.c
	(_dbus_decompose_path): fix to handle path "/" properly
	(run_decompose_tests): add tests for path decomposition

	* glib/dbus-gutils.c (_dbus_gutils_split_path): fix to handle "/"
	properly

	* glib/dbus-gobject.c (handle_introspect): fix quotes

	* test/glib/run-test.sh: support launching the bus, then running
	dbus-viewer

	* test/glib/test-service-glib.c (main): put in a trivial gobject
	subclass and register it on the connection

	* bus/driver.c (bus_driver_handle_introspect): implement
	introspection of the bus driver service

	* dbus/dbus-protocol.h: add #defines for the XML namespace,
	identifiers, doctype decl

	* bus/driver.c (bus_driver_handle_get_service_owner): handle
	attempts to get owner of DBUS_SERVICE_ORG_FREEDESKTOP_DBUS by
	returning the service unchanged.
	(bus_driver_handle_message): remove old check for reply_serial in
	method calls, now the message type deals with that
	(bus_driver_handle_message): handle NULL interface

	* glib/dbus-gproxy.c (dbus_g_proxy_get_bus_name): new function

	* glib/dbus-gloader-expat.c (description_load_from_string): allow
	-1 for len

	* tools/dbus-viewer.c: add support for introspecting a service on
	a bus

	* glib/dbus-gproxy.c (dbus_g_pending_call_ref): add
	(dbus_g_pending_call_unref): add
2005-01-30 05:18:44 +00:00
Havoc Pennington
2922b0e88b 2005-01-29 Havoc Pennington <hp@redhat.com>
* tools/dbus-tree-view.c: add support for displaying properties.
	(run dbus-viewer with an introspect xml file as arg, then resize
	the window so the tree elements show up, not sure what that is)

	* glib/dbus-gobject.c (handle_introspect): return
	org.freedesktop.Properties and org.freedesktop.Introspectable
	interfaces when we are introspected.

	* doc/dbus-specification.xml: allow empty interface name when
	Get/Set a property
2005-01-29 20:12:22 +00:00
Havoc Pennington
fd3e49f249 2005-01-29 Havoc Pennington <hp@redhat.com>
* glib/Makefile.am: rename dbus-glib-tool to dbus-binding-tool;
	though it uses glib, it could be extended for any binding in
	principle

	* glib/dbus-gobject.c (gobject_message_function): change to the
	new way properties work

	* dbus/dbus-protocol.h: add the new interfaces

	* doc/dbus-specification.xml: document the introspection format,
	Introspectable interface, and add an org.freedesktop.Properties
	interface.

	* glib/dbus-gparser.c: add support for a <property> element

	* glib/dbus-gidl.c: add PropertyInfo

	* glib/dbus-gobject.c (handle_introspect): put the outermost
	<node> outside the signal and property descriptions.
	(introspect_properties): export properties as <property> rather
	than as method calls
2005-01-29 19:52:19 +00:00
Havoc Pennington
602c4b05c4 2005-01-28 Havoc Pennington <hp@redhat.com>
* doc/TODO, doc/dbus-specification.xml: spec and TODO tweaks
	related to authentication protocol
2005-01-29 04:21:32 +00:00
John (J5) Palmieri
216fa619f3 * python/dbus_bindings.pyx.in: Updated to handle new D-BUS type system
- BUS_ACTIVATION -> BUS_STARTER
	- DBUS_BUS_ACTIVATION -> DBUS_BUS_STARTER
	- class MessageIter (__init__): Added recursion checking
	so we throw a nice error instead of just disconnecting from the
	bus.
	(get): Added arg_type parameter for recursion.
	Removed the nil type
	Added signiture type placeholder (not implemented)
	Added struct type placeholder (not implemented)
	Added varient type placeholder (not implemented)
	Commented out dict type for now
	(get_element_type): renamed from get_array_type
	(get_*): changed to use the dbus_message_iter_get_basic API
	(get_*_array): removed in favor of recursive get_array method
	(get_array): new recursive method which calls get to marshal
        the elements of the array
	(value_to_dbus_sig): New method returns the corrasponding
	dbus signiture to a python value
	(append): Comment out dict handling for now
	Handle lists with the new recursive API
	Comment out None handling for now
	(append_nil): removed
	(append_*): changed to use dbus_message_iter_append_basic API
	(append_*_array): removed in favor of recursive append_array
	method
	(__str__): Make it easier to print out recursive iterators
	for debugging
	- class Message (__str__): moved type inspection to the
	MessageIter class' __str__ method
	(get_iter): Added an append parameter wich defaults to False
	If True use the new API's to create an append iterator

* python/dbus.py: Update to use new bindings API
	- TYPE_ACTIVATION -> TYPE_STARTER
	- class Bus (_get_match_rule): GetServiceOwner -> GetNameOwner
	- class ActivationBus -> class StarterBus
	- class RemoteObject (__call__): get an append iterator
	- (_dispatch_dbus_method_call): get an append iterator
	- class Object (emit_signal): get an append iterator

* python/examples/: Fixed up the examples to work with the new API
2005-01-28 19:09:55 +00:00
Joe Shaw
cbe79dee56 2005-01-28 Joe Shaw <joeshaw@novell.com>
* configure.in: Bump version up to 0.30.

	* HACKING: Add a release item to bump the version number up after
	a release.
2005-01-28 17:04:35 +00:00
Havoc Pennington
bb98998d68 small spec fix 2005-01-28 15:02:41 +00:00
Havoc Pennington
c7816d45e8 2005-01-28 Havoc Pennington <hp@redhat.com>
* doc/dbus-specification.xml: update to describe 16-bit types and
	dict entries

	* dbus/dbus-marshal-basic.c (_dbus_unpack_uint16): fix broken
	assertion

	* dbus/dbus-protocol.h (DBUS_TYPE_DICT_ENTRY): add DICT_ENTRY as a
	type

	* dbus/dbus-marshal-recursive.c: implement
2005-01-28 05:30:53 +00:00
Havoc Pennington
3ed9db546e 2005-01-27 Havoc Pennington <hp@redhat.com>
* dbus/dbus-arch-deps.h.in: add 16/32-bit types

	* configure.in: find the right type for 16 and 32 bit ints as well
	as 64

	* dbus/dbus-protocol.h (DBUS_TYPE_INT16, DBUS_TYPE_UINT16): add
	the 16-bit types so people don't have to stuff them in 32-bit or
	byte arrays.
2005-01-28 03:06:56 +00:00
Havoc Pennington
4506b6594b remove completed byteswap item 2005-01-28 00:08:28 +00:00
Havoc Pennington
fddbc09c4a 2005-01-27 Havoc Pennington <hp@redhat.com>
* dbus/dbus-message.c: byteswap the message if you init an
	iterator to read/write from it

	* dbus/dbus-marshal-byteswap.c: new file implementing
	_dbus_marshal_byteswap()

	* dbus/dbus-marshal-basic.c: add _dbus_swap_array()
2005-01-27 23:39:26 +00:00
Havoc Pennington
382d5ad0b4 2005-01-26 Havoc Pennington <hp@redhat.com>
* dbus/dbus-marshal-validate-util.c: break this out (and fix
	build, apparently - nobody noticed?)
2005-01-27 01:50:01 +00:00