[PATCH 2/5] server: Keep children belonging to other threads alive in destroy_thread_windows.

[Jacek Caban]

Signed-off-by: Jacek Caban 
---
Currently when a thread exits, we destroy such windows on server side, 
but WND pointer on client side is alive and the right thread is notified 
to do a proper clean up. Until clean up is done, we have an inconsistent 
state when some functions will work on child handle (actually, enough of 
them to do the clean up), but ones that require a server call fail (as 
shown in the test from previous patch, using SetWindowLong as an example).

dlls/user32/tests/msg.c |  1 -
  server/window.c         | 33 ++++++++++++++++++++++++++++++---
  2 files changed, 30 insertions(+), 4 deletions(-)

[Marvin]

Hi,

While running your changed tests, I think I found new failures.
Being a bot and all I'm not very good at pattern recognition, so I might be
wrong, but could you please double-check?

Full results can be found at:
https://testbot.winehq.org/JobDetails.pl?Key=106730

Your paranoid android.

=== w8 (32 bit report) ===

user32:
msg.c:15294: Test failed: unexpected message 31f
msg.c:15295: Test failed: bad wparam 1
msg.c:15301: Test failed: unicode WM_CHAR: 0: the msg sequence is not complete: expected 0102 - actual 0000