There is one test failure in winWm.test on Windows.
winWm.test
==== winWm-5.1 UpdateGeometryInfo: menu
resizing FAILED
==== Contents of test case:
set result {}
toplevel .t
frame .t.f -width 150 -height 50 -bg red
pack .t.f
update
set result [winfo height .t]
menu .t.m
.t.m add command -label foo
.t conf -menu .t.m
update
lappend result [winfo height .t]
.t.m add command -label "thisisreallylong"
.t.m add command -label "thisisreallylong"
update
lappend result [winfo height .t]
destroy .t
set result
---- Result was:
50 50 32
---- Result should have been (exact matching):
50 50 50
==== winWm-5.1 FAILED
Logged In: YES
user_id=32170
This discrepancy is quite large, and seems to be due to
the native menu bar wrapping around onto a second
row, and this, for some reason causes the window to
shrink. All very odd.
Logged In: YES
user_id=202636
I assume that this test is not valid for Windows native
menus because they wrap the menubar. Perhaps we can add a
unixOnly constraint to this one?
Logged In: YES
user_id=596509
Tk: v8.5a2 HEAD at 30-08-04
OS: WinXPsp1
Built with makefile.vc using VC7
==== winWm-5.1 UpdateGeometryInfo: menu resizing FAILED
==== Contents of test case:
set result {}
toplevel .t
frame .t.f -width 150 -height 50 -bg red
pack .t.f
update
set result [winfo height .t]
menu .t.m
.t.m add command -label foo
.t conf -menu .t.m
update
lappend result [winfo height .t]
.t.m add command -label "thisisreallylong"
.t.m add command -label "thisisreallylong"
update
lappend result [winfo height .t]
destroy .t
set result
---- Result was:
50 50 31
---- Result should have been (exact matching):
50 50 50
==== winWm-5.1 FAILED
Logged In: YES
user_id=202636
The relevant code in tkWinWm.c:5683 suggests that we are
making attempts to keep resizing the window until we can get
the requested size.
This is evidently no longer working but suggests to me that
this is _not_ an invalid test but that something has changed
in Windows window management at some point.
Assigning to patthoyts since he seems to know what's going on here.