jhsieh@grapevine.EBay.Sun.COM (James Hsieh) (07/13/90)
Hi,
Could you help me to understand the propagation of device events?
Assuming window tree hierarchy in the server is:
root
/ | \
c1 A B
/
c2
and
pointer is in c2,
button pressed,
no active/passive grabs,
no do-not-propagate masks,
just normal event distribution in the server.
the screen appearance is:
---------------------------
| root |
| ----------------- |
| | A | |
| -------------- | |
| | B | | |
| | |----- |
| | ---- |c1 | |
| | |c2| | | |
| | |x | | | |
| | ---- | | |
| -------------- | |
| | -------------- |
| | | |
| ----------------- |
| |
---------------------------
case 1: client 1 selects ButtonPressMask on window A
client 2 selects ButtonPressMask on window B
client 3 selects ButtonPressMask on window c1, but not c2
Question:
1. what is the seach path for this ButtonPress event?
2. which client will receive this ButtonPress event?
3. is the search path from source window to event window
determined by looking up the window hierarchy in the server,
not by the current appearance on screen?
My answer is:
1. c2-c1-stop.
2. server sends this event to client 3.
3. yes. the event propagation is based on window hierarchy IN
the server, NOT based on physical layout of screen.
case 2: client 1 selects ButtonPressMask on window A
client 2 selects ButtonPressMask on window B
client 3 is not interested in ButtonPress on c1 or c2.
Question:
1. what is the seach path for this ButtonPress event?
2. which client will receive this button press event?
3. why isn't window B included in the search path
to receive the event?
My answer is:
1. c2-c1-root-stop.
2. evnet thrown away, none of the clients receives this event.
3. window B does NOT contain the pointer, so window B
is not in the search path to receive this evnet?
case 3: none of the clients selects ButtonPressMask on any window.
Question:
1. what is the search path for this ButtonPress event?
2. which client will receive this button press event?
My answer is:
1. search path is still c2-c1-root-Stop
2. event thrown away, none of the clients receives this event.
Thank you very much for your kind help.
Yungchien rws@EXPO.LCS.MIT.EDU (Bob Scheifler) (07/14/90)
On a quick scan I think your statements are correct, but your picture looks wrong. It would appear to show c1 underneath B, but c2 on top of B. If c2 is a child of c1 then this it not possible in X.
mouse@SHAMASH.MCRCIM.MCGILL.EDU (der Mouse) (07/14/90)
> Assuming window tree hierarchy in the server is: > root > / | \ > c1 A B > / > c2 > and > pointer is in c2, > button pressed, > no active/passive grabs, > no do-not-propagate masks, > just normal event distribution in the server. Well, grabs and do-not-propagate masks are part of normal event distribution, but aside from that.... > the screen appearance is: > --------------------------- > | root | > | ----------------- | > | | A | | > | -------------- | | > | | B | | | > | | |----- | > | | ---- |c1 | | > | | |c2| | | | > | | |x | | | | > | | ---- | | | > | -------------- | | > | | -------------- | > | | | | > | ----------------- | > | | > --------------------------- This is not possible. Given your hierarchy above, c2 can't be in front of B unless c1 also is. (All the children of a window W always have the same relationship as W in the global front-to-back stacking order relative to any other window X which is neither W nor a child of W. In this case, W is c1, the child is c2, and X is B.) If, however, c1 in your picture is raised to in front of B, all the answers you gave to your own questions are (according to my own understanding of event distribution, which may well be flawed) correct. der Mouse old: mcgill-vision!mouse new: mouse@larry.mcrcim.mcgill.edu