4 USER implements windowing and messaging subsystems. It also
5 contains code for common controls and for other miscellaneous
6 stuff (rectangles, clipboard, WNet, etc). Wine USER code is
7 located in windows/, controls/, and misc/ directories.
11 Windows are arranged into parent/child hierarchy with one
12 common ancestor for all windows (desktop window). Each window
13 structure contains a pointer to the immediate ancestor (parent
14 window if WS_CHILD style bit is set), a pointer to the sibling
15 (returned by GetWindow(..., GW_NEXT)), a pointer to the owner
16 window (set only for popup window if it was created with valid
17 hwndParent parameter), and a pointer to the first child
18 window (GetWindow(.., GW_CHILD)). All popup and non-child windows
19 are therefore placed in the first level of this hierarchy and their
20 ancestor link (wnd->parent) points to the desktop window.
22 Desktop window - root window
25 popup -> wnd1 -> wnd2 - top level windows
28 child1 child2 -> child3 child4 - child windows
30 Horizontal arrows denote sibling relationship, vertical lines
31 - ancestor/child. To summarize, all windows with the same immediate
32 ancestor are sibling windows, all windows which do not have desktop
33 as their immediate ancestor are child windows. Popup windows behave
34 as topmost top-level windows unless they are owned. In this case the
35 only requirement is that they must precede their owners in the top-level
36 sibling list (they are not topmost). Child windows are confined to the
37 client area of their parent windows (client area is where window gets
38 to do its own drawing, non-client area consists of caption, menu, borders,
39 intrinsic scrollbars, and minimize/maximize/close buttons).
41 Another fairly important concept is "z-order". It is derived from
42 the ancestor/child hierarchy and is used to determine "above/below"
43 relationship. For instance, in the example above, z-order is
44 child1->popup->child2->child3->wnd1->child4->wnd2->desktop. Current
45 active window ("foreground window" in Win32) is moved to the front
46 of z-order unless its top-level ancestor owns popup windows.
48 All these issues are dealt with (or supposed to be) in
51 Wine specifics: in default and managed mode each top-level window
52 gets its own X counterpart with desktop window being basically a
53 fake stub. In desktop mode, however, only desktop window has X
54 window associated with it.
56 2. Messaging subsystem
58 Each Windows task/thread has its own message queue - this is where
59 it gets messages from. Messages can be generated on the fly
60 (WM_PAINT, WM_NCPAINT, WM_TIMER), they can be created by the system
61 (hardware messages), they can be posted by other tasks/threads
62 (PostMessage), or they can be sent by other tasks/threads (SendMessage).
66 First the system looks for sent messages, then for posted messages,
67 then for hardware messages, then it checks if the queue has the
68 "dirty window" bit set, and, finally, it checks for expired
69 timers. See windows/message.c.
71 From all these different types of messages, only posted messages go
72 directly into the private message queue. System messages (even in
73 Win95) are first collected in the system message queue and then
74 they either sit there until Get/PeekMessage gets to process them
75 or, as in Win95, if system queue is getting clobbered, a special
76 thread ("raw input thread") assigns them to the private
77 queues. Sent messages are queued separately and the sender sleeps
78 until it gets a reply. Special messages are generated on the fly
79 depending on the window/queue state. If the window update region is
80 not empty, the system sets the QS_PAINT bit in the owning queue and
81 eventually this window receives a WM_PAINT message (WM_NCPAINT too
82 if the update region intersects with the non-client area). A timer
83 event is raised when one of the queue timers expire. Depending on
84 the timer parameters DispatchMessage either calls the callback
85 function or the window procedure. If there are no messages pending
86 the task/thread sleeps until messages appear.
88 There are several tricky moments (open for discussion) -
90 a) System message order has to be honored and messages should be
91 processed within correct task/thread context. Therefore when
92 Get/PeekMessage encounters unassigned system message and this
93 message appears not to be for the current task/thread it should
94 either skip it (or get rid of it by moving it into the private
95 message queue of the target task/thread - Win95, AFAIK) and
96 look further or roll back and then yield until this message
97 gets processed when system switches to the correct context
98 (Win16). In the first case we lose correct message ordering, in
99 the second case we have the infamous synchronous system message
100 queue. Here is a post to one of the OS/2 newsgroup I found to
103 " Here's the problem in a nutshell, and there is no good solution.
104 Every possible solution creates a different problem.
106 With a windowing system, events can go to many different windows.
107 Most are sent by applications or by the OS when things relating to
108 that window happen (like repainting, timers, etc.)
110 Mouse input events go to the window you click on (unless some window
113 So far, no problem. Whenever an event happens, you put a message on
114 the target window's message queue. Every process has a message
115 queue. If the process queue fills up, the messages back up onto the
118 This is the first cause of apps hanging the GUI. If an app doesn't
119 handle messages and they back up into the system queue, other apps
120 can't get any more messages. The reason is that the next message in
121 line can't go anywhere, and the system won't skip over it.
123 This can be fixed by making apps have bigger private message queues.
124 The SIQ fix does this. PMQSIZE does this for systems without the SIQ
125 fix. Applications can also request large queues on their own.
127 Another source of the problem, however, happens when you include
128 keyboard events. When you press a key, there's no easy way to know
129 what window the keystroke message should be delivered to.
131 Most windowing systems use a concept known as "focus". The window
132 with focus gets all incoming keyboard messages. Focus can be changed
133 from window to window by apps or by users clicking on winodws.
135 This is the second source of the problem. Suppose window A has focus.
136 You click on window B and start typing before the window gets focus.
137 Where should the keystrokes go? On the one hand, they should go to A
138 until the focus actually changes to B. On the other hand, you
139 probably want the keystrokes to go to B, since you clicked there
142 OS/2's solution is that when a focus-changing event happens (like
143 clicking on a window), OS/2 holds all messages in the system queue
144 until the focus change actually happens. This way, subsequent
145 keystrokes go to the window you clicked on, even if it takes a while
146 for that window to get focus.
148 The downside is that if the window takes a real long time to get focus
149 (maybe it's not handling events, or maybe the window losing focus
150 isn't handling events), everything backs up in the system queue and
151 the system appears hung.
153 There are a few solutions to this problem.
155 One is to make focus policy asynchronous. That is, focus changing has
156 absolutely nothing to do with the keyboard. If you click on a window
157 and start typing before the focus actually changes, the keystrokes go
158 to the first window until focus changes, then they go to the second.
159 This is what X-windows does.
161 Another is what NT does. When focus changes, keyboard events are held
162 in the system message queue, but other events are allowed through.
163 This is "asynchronous" because the messages in the system queue are
164 delivered to the application queues in a different order from that
165 with which they were posted. If a bad app won't handle the "lose
166 focus" message, it's of no consequence - the app receiving focus will
167 get its "gain focus" message, and the keystrokes will go to it.
169 The NT solution also takes care of the application queue filling up
170 problem. Since the system delivers messages asynchronously, messages
171 waiting in the system queue will just sit there and the rest of the
172 messages will be delivered to their apps.
174 The OS/2 SIQ solution is this: When a focus-changing event happens,
175 in addition to blocking further messages from the application queues,
176 a timer is started. When the timer goes off, if the focus change has
177 not yet happened, the bad app has its focus taken away and all
178 messages targetted at that window are skipped. When the bad app
179 finally handles the focus change message, OS/2 will detect this and
180 stop skipping its messages.
183 As for the pros and cons:
185 The X-windows solution is probably the easiest. The problem is that
186 users generally don't like having to wait for the focus to change
187 before they start typing. On many occasions, you can type and the
188 characters end up in the wrong window because something (usually heavy
189 system load) is preventing the focus change from happening in a timely
192 The NT solution seems pretty nice, but making the system message queue
193 asynchronous can cause similar problems to the X-windows problem.
194 Since messages can be delivered out of order, programs must not assume
195 that two messages posted in a particular order will be delivered in
196 that same order. This can break legacy apps, but since Win32 always
197 had an asynchronous queue, it is fair to simply tell app designers
198 "don't do that". It's harder to tell app designers something like
199 that on OS/2 - they'll complain "you changed the rules and our apps
202 The OS/2 solution's problem is that nothing happens until you try to
203 change window focus, and then wait for the timeout. Until then, the
204 bad app is not detected and nothing is done." (by David Charlap)
207 b) Intertask/interthread SendMessage. The system has to inform the
208 target queue about the forthcoming message, then it has to
209 carry out the context switch and wait until the result is
210 available. In Win16 it is done by putting necessary parameters
211 into the queue structure and do a DirectedYield() call.
212 However, in Win32 there could be several messages pending sent
213 by preemptively executing threads, and in this case SendMessage
214 has to build some sort of message queue for sent
215 messages. Another issue is what to do with messages sent to the
216 sender when it is blocked inside its own SendMessage. At this
217 point Wine does not address any of these problems.