| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
| |
* win/configure.in: Don't run the AC_CYGWIN
macro since it uses AC_CANONICAL_HOST under
autoconf 2.5X. Just check to see if __CYGWIN__
is defined by the compiler and set the
ac_cv_cygwin variable based on that.
[Tcl Bug 705912]
|
|
|
|
| |
commit. Thanks Larry Virden. [Bug 705567]
|
|
|
|
|
|
|
|
| |
* macosx/Makefile:
* macosx/README: added support for giving 'make' the location
of Tcl.framework and tclsh to build and link against.
* macosx/tkMacOSXXStubs.c: fixed crash in [winfo server].
|
|
|
|
|
|
| |
to run only on Unix. Alternative approach did not match maintainer
habits.
* tests/spinbox.test (spinbox-17.4): Avoid long failure message.
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
for a detailed explanation of why this is good.
This will no doubt break the build on some
platforms, let the flaming begin.
* unix/configure: Regen with autoconf 2.57.
* unix/configure.in: Require autoconf 2.57.
* unix/tcl.m4: Update from Tcl.
* win/configure: Regen with autoconf 2.57.
* win/configure.in: Require autoconf 2.57.
* win/tcl.m4: Update from Tcl.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
* macosx/tkMacOSXMenu.c (EventuallyInvokeMenu): New function,
used to invoke menu commands at idle time.
(TkMacOSXDispatchMenuEvent): Don't immediately dispatch menu
commands, wait till the idle loop to do so. This is more like
what is done on Windows, and avoids the crash from destroying
a menu in it's command.
* macosx/tkMacOSXHLEvents.c (ReallyKillMe): Don't force the shell
to exit. According to the OS X HI guidelines, it should be
possible to cancel an attempt to quit, and if we force the kill,
here, it would not be possible to implement this.
|
|
|
|
|
|
| |
constraint to tests that depend on the
testwrapper command. These tests should
not be run under Windows for example.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
that the resizable command keeps scripts from
changing the size of windows since it is
not true. The resizable command only applies
to user sizing via user interaction.
* tests/unixWm.test: Replace broken tests with
the nonPortable constraint with new tests for
maxsize and minsize options. These tests
verify that setting the minsize and maxsize
will resize the window if needed, and that
the wm hints will be updated with the new sizes.
* tests/wm.test: Add an exhaustive set of tests
for the wm maxsize and wm minsize commands.
These tests verify that setting the minsize
and maxsize will resize the window if needed.
These tests have only been run under Win98
and Window Maker under Linux, so further tweaking
may be needed for other systems.
* unix/tkUnixWm.c (UpdateGeometryInfo, UpdateSizeHints):
Fixup comments and initialization for the minWidth,
minHeight, maxWidth, maxHeight, width, and height
members of the WmInfo struct. Check to ensure
that a new toplevel window size is not larger
than the maxsize or smaller than the minsize
when updating the geometry at idle time.
Pass new width and height values to the
UpdateSizeHints method so that it can properly
set the window min and max sizes for a window
that cannot be resized by the user. This fixes
a bug where the window resizes back to the original
size when the user clicks on the window border.
* win/tkWinWm.c (UpdateGeometryInfo):
Fixup comments and initialization for the minWidth,
minHeight, maxWidth, maxHeight, width, and height
members of the WmInfo struct. Check to ensure
that a new toplevel window size is not larger
than the maxsize or smaller than the minsize
when updating the geometry at idle time.
[Patch 568861]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Check for a NULL masterPtr and slavePtr in the
GridStructureProc code to ensure that a Gridder
created before some error condition is ignored
when it comes to geometry calculations. This
approach closely matches the pack implementation.
Keep track of a -in argument to a grid command
in order to detect the case of an already
gridded widget that wants to change some options.
The previous implementation could make repeated
and unnecessary calls to Tk_ManageGeometry.
Replace use of "parent" with "master" in comments
throughout the file.
* generic/tkPack.c (PackStructureProc): Check for
a NULL masterPtr before other checks so that a
slave created under certain error conditions
is cleaned up properly.
Replace use of "parent" with "master" in comments
throughout the file.
* generic/tkPlace.c (CreateSlave, ConfigureSlave,
SlaveStructureProc):
Don't call Tk_ManageGeometry in CreateSlave since
this was causing incorrect results in some error
cases. Rework the ConfigureSlave method so that
slave setup is done in one place. The call to
Tk_ManageGeometry was added to the one place
where a slave is setup. When a slave is configured
but the master is not changed, simply goto the
scheduleLayout label. Check for a NULL master
in SlaveStructureProc for the sake of readability.
* tests/grid.test:
* tests/pack.test:
* tests/place.test: Add test to check that a
winfo manager call does not return incorrect
results after an error condition is hit.
[Patch 693063]
|
|
|
|
| |
of the install step.
|
|
|
|
|
| |
MacOS 9 behavior, which was to insert a separator item after the system help
menus, but this doesn't happen on X...
|
| |
|
|
|
|
|
| |
they could never be done portably as they'd inevitably depend on the total
amount of memory available to the process... :^(
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
* generic/tk.h: Tk to 8.5a0.
* library/tk.tcl:
* mac/README:
* macosx/Wish.pbproj/project.pbxproj:
* unix/README:
* unix/configure.in:
* unix/tk.spec:
* win/README:
* win/configure.in:
* win/makefile.vc:
* win/tcl.m4:
* unix/configure: autoconf
* win/configure:
|
| |
|
|
|
|
| |
underlying field to prevent a crash on Windows. [Bug #612110]
|
| |
|
|
|
|
|
|
| |
* win/tkWinDraw.c:
* win/tkWinFont.c (Tk_DrawChars): add support for simple XOR text
drawing on Windows. [Patch #685388] (martin)
|
|
|
|
| |
menusInitialized on finalize. [Bug #548729]
|
|
|
|
|
|
|
|
|
|
|
|
| |
parameters to posix style path names.
[Bug #621672] (steffen, Benjamin Riefenstahl)
* macosx/tkMacOSXDialog.c:
* macosx/tkMacOSXWm.c:
* macosx/Wish.pbproj/project.pbxproj:
* macosx/tkMacOSXUtil.c (removed):
* macosx/tkMacOSXUtil.h (removed): removed all references to
tkMacOSXUtil.{c,h}, made obsolete by the patch above. [Bug #621672]
|
|
|
|
| |
first best font match in user font path. [Bug #647497] (dal zotto)
|
|
|
|
|
| |
* tests/listbox.test: Allow 'selection includes' to respond when
disabled (but only 'includes'). [Bug #632514]
|
|
|
|
|
| |
of disabled check/radiobutton to be more distinctive on unix.
* tests/unixButton.test: [Bug #669595] (hintermayer)
|
|
|
|
|
|
|
| |
* tests/entry.test: return 1 if selection is present even if
* tests/spinbox.test: entry/spinbox is disabled, as selection get
will still return the selection (although selection still ignore
modify requests when entry/spinbox is disabled). [Bug #637828]
|
|
|
|
|
|
| |
* tkMacOSXDialog.c (NavServicesGetFile): Fix bug with
kNavCtlSetSelection - control data should be a pointer
not a handle.
|
|
|
|
|
|
| |
* library/msgbox.tcl (MessageBox): Yet more places that did
complex and fragile width calculations instead of using grid,
though with these it is the -uniform option that saves the day.
|
| |
|
|
|
|
|
| |
long a file type item description string can get due to appending
of types. [Bug #617392]
|
| |
|
|
|
|
|
|
|
|
|
|
| |
binding for corresponding Quit menu item shortcut.
* macosx/README: updated instructions for embedded build,
added various improvements by Russell Owen.
* unix/tcl.m4: update from Tcl.
* unix/configure: regen.
|
|
|
|
|
| |
"disable" scrollbar when on Windows when there is nothing to
scroll. This is Windows style, and fixes [Bug #624116].
|
|
|
|
| |
size when increasing the size of the panedwindow. [Bug #689099]
|
|
|
|
| |
It served no useful purpose whatsoever...
|
|
|
|
| |
test would be very platform- and build-specific. [Bugs #618872 and #689357]
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
| |
* tkMacOSXMouseEvents.c (GeneratePollingEvents): In the case
where there was a capture window, we were sending the events to
the capture window. But the capture window (return value or
TkMacOSXGetCapture) is always a toplevel. So this is wrong in
the case that the Event's toplevel IS the capture window - in
which case the event should go to the subwindow most closely
containing the event.
Fixes bug #688188
|
|
|
|
|
|
|
|
|
|
| |
two parts of the patches in Patch Tracker #622582 -
new-evthdlng.2003-02-12.diff and basic-keyboard.2003-02-10.diff. The
second part puts translation of MacOS X keycodes to characters on a
better footing. The first part relaxs Tk's policy of consuming all
events unless it can see they go to windows it didn't create. This
change gets the little traffic lights working, and should make things
like QuickTimeTcl easier to implement.
|
|
|
|
|
|
| |
* doc/text.n: Update double and triple click
docs to match implementation changes made on
2002-02-14.
|
| |
|
| |
|
|
|
|
|
|
| |
* macosx/Makefile: reworked embedded build to no longer require
relinking but to use install_name_tool instead to change the
install_names for embedded frameworks.
|
| |
|
|
|
|
| |
* unix/tcl.m4: Update from Tcl (HP xnet [Bug #651811]).
|
|
|
|
| |
* unix/tcl.m4: Update from Tcl.
|
| |
|
|
|
|
| |
well as entries when tabbing in. [Patch #683813] (bron)
|
|
|
|
| |
sub on embedded newlines. [Patch #688572] (bonfield)
|
| |
|