| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
| |
quoting introduced by [Patch 521356] causing the installed to
fail.
|
| |
|
|
|
|
|
| |
match the <eol> of the file itself. Windows users where getting
\r\r\n. Problem fixed.
|
| |
|
| |
|
| |
|
|
|
|
| |
pkgIndex.tcl file to enable Tk as a loadable package [Patch 521356]
|
|
|
|
|
|
|
|
|
|
|
| |
* unix/configure.in: Added support for symlinks and compression
* unix/Makefile.in: when installing the manpages. [Patch 518052]
Default is still hardlinks and no compression.
* unix/mkLinks: generated
* unix/configure:
* unix/README: Added documentation for the new features.
|
| |
|
|
|
|
| |
* unix/tcl.m4 (SC_PATH_TCLCONFIG): by ${libdir}.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
and already mapped toplevel are raised and receive
the focus when deiconified.
* tests/wm.test: Add wm deiconify tests. Check that
a toplevel that has never been mapped is not mapped
by the deiconify command since it should be done
at idle by MapFrame.
* win/tkWinWm.c (Tk_WmCmd): Check the WM_NEVER_MAPPED
flag while processing the wm deiconify command.
The WM_UPDATE_PENDING flag should never be set when
WM_NEVER_MAPPED is set, but double check so that
the implementation is more explicit and matches
the comment just above.
Return without invoking TkWmRestackToplevel or
TkSetFocusWin on a toplevel that has never been
mapped. This fixes a bug where a toplevel is mapped
with the wrong size and is then resized by the
idle call to MapFrame. [Tk bug 233150]
|
| |
|
| |
|
|
|
|
|
|
|
| |
wrapper introduced in r1.41 on 2002-06-15 because it prevented
geometry setting from taking effect if the window was not on the
screen. Another check may go in it's place as IsIconic and
IsZoomed should not be passed NULL.
|
|
|
|
|
|
|
|
|
|
| |
calls to wm transient tests so that idle handlers get
run. This is needed to get the tests to pass under Win32.
* unix/tkUnixWm.c (WmWaitMapProc): Move the special
transient withdrawn check into the if body to
make it easier to set a breakpoint on this test
inside a debugger. No functional changes.
* win/tkWinWm.c (WmWaitVisibilityOrMapProc): Ditto.
|
|
|
|
|
|
|
|
| |
* mac/tkMacWm.c (Tk_WmCmd): docs with mac and unix stubs.
* unix/tkUnixWm.c (Tk_WmCmd):
* win/tkWinWm.c (Tk_WmCmd):
* tests/unixWm.test:
* tests/winWm.test: more wm attr tests will be needed.
|
| |
|
|
|
|
| |
(may be when embedded). [Bug #548791] (halliday)
|
|
|
|
| |
to an assert, and wrapped panic in #ifdef PURIFY. [Bug #568701]
|
|
|
|
| |
on Windows. [Bug #464451] (darley)
|
|
|
|
| |
[Bug #553207] and made Tab a default expansion key (like Escape).
|
|
|
|
| |
containerPtr. The core of this bug is likely elsewhere. [Bug #476176]
|
|
|
|
|
|
|
| |
* generic/tkText.c (TextWidgetCmd): enhances the text get and
* generic/tkTextIndex.c (TkTextGetIndex): delete methods to accept
* tests/text.test: multiple range pairs.
This handles the delete case in an atomic, fixed-index fashion.
|
|
|
|
|
|
|
|
|
|
|
| |
transient does not get remapped by state changes
in the master.
* unix/tkUnixWm.c (Tk_WmCmd, WmWaitMapProc):
* win/tkWinWm.c (Tk_WmCmd, WmWaitVisibilityOrMapProc):
Add a WM_TRANSIENT_WITHDRAWN flag that gets set by the
withdraw, deiconify, or state wm subcommands. Check
this flag before mapping a transient when processing
a MapNotify event. [Tk bug 570764]
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
* generic/tkText.c: undo/redo stack to not be tied solely to the
* generic/tkText.h: text widget. The APIs are still private.
* generic/tkUndo.c: This also adds a stack limiting ability and
* generic/tkUndo.h: a -maxundo option to the text widget (in
* library/text.tcl: addition to the options from TIP #26) should
* mac/tkMacDefault.h: users want to limit the undo/redo stack
* tests/text.test: (should not be necessary in most cases).
* unix/Makefile.in: [Patch #554763] (callewart)
* unix/tkUnixDefault.h:
* win/Makefile.in:
* win/makefile.vc:
* win/tkWinDefault.h:
|
| |
|
|
|
|
|
| |
executable on Unix. If there are similar dependencies on other
platforms, they can probably be removed as well. [Bug 572134].
|
|
|
|
|
|
|
|
| |
* generic/tkListbox.c (DisplayListbox):
* mac/tkMacDefault.h: TIP #94 implementation adding -activestyle
* tests/listbox.test: option to the listbox. This adds the ability
* unix/tkUnixDefault.h: to have listboxes look native on Windows, and
* win/tkWinDefault.h: "nicer" elsewhere using the 'dotbox' style.
|
| |
|
|
|
|
|
| |
* generic/tkPanedWindow.c: ensure that sash index is lower bounds
checked. [Bug #548727]
|
|
|
|
|
|
| |
to set dispPtr->clipWindow to NULL, this was
accidently removed by last commit. Fixes
a crash while running the tests under win32.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Handle case where Tk_DestroyWindow is invoked
on clipboard and send windows.
* generic/tkClipboard.c (TkClipCleanup):
Invoke Tk_DestroyWindow to cleanup the
dispPtr->clipWindow. Call Tcl_Preserve
and Tcl_Release on the window to avoid an
invalid memory ref on shutdown.
* generic/tkEvent.c (Tk_HandleEvent):
Panic if XCreateIC is invoked twice for
the same window. This should never happen,
the check were just added to make sure it
does not since this could lead to crashes
in XCloseIM.
* generic/tkFocus.c (TkFocusDeadWindow):
Handle case where Tk_DestroyWindow is invoked
on clipboard and send windows.
* generic/tkOption.c (TkOptionDeadWindow): Ditto.
* generic/tkWindow.c (TkCloseDisplay): Move
deletion of dispPtr->winTable after TkpCloseDisplay
call since Tk_DestroyWindow uses it and could
be called by TkpCloseDisplay for clipboard/send windows.
Also invoke ckfree for the dispPtr instead of
doing it in TkpCloseDisplay.
(Tk_DestroyWindow): Check for a null winPtr->mainPtr
before doing certain cleanup tasks so the we can
invoke Tk_DestroyWindow on clipboard and send windows.
We need to do this so that XDestroyIC will get invoked
for the input contexts of each window.
* mac/tkMacXStubs.c (TkpCloseDisplay): Don't free
the displayPtr since this is now done in TkCloseDisplay.
* unix/tkUnixEvent.c (TkpCloseDisplay, OpenIM): Remove
conditional compilation around calls to XCloseIM
since I am confident that the crashes related to
input contexts has been fixed. Don't free
the displayPtr since this is now done in TkCloseDisplay.
* unix/tkUnixSend.c (TkSendCleanup): Invoke the
Tk_DestroyWindow method to cleanup the special
send window. This will call XDestroyIC and thereby
avoid a crash in XCloseIM. The send window needs
to be Tcl_Preserve and Tcl_Release to avoid an
invalid memory ref on shutdown.
* win/tkWinX.c (TkpCloseDisplay): Don't free
the displayPtr since this is now done in TkCloseDisplay.
|
| |
|
|
|
|
| |
Frederic Bonnet to supply docs+tests as soon as possible.
|
| |
|
|
|
|
|
|
| |
Tcl_Preserve and Tcl_Release for the masterPtr->winPtr
window to avoid accessing memory that had already
been deallocated in DeleteImage.
|
|
|
|
|
|
|
|
|
|
| |
scripts from Tcl's accepted FR #565088.
* generic/tk.h: Changed RESOURCE_INCLUDED to be RC_INVOKED
as the RC tool defines this already by default.
* win/rc/tk.rc:
* win/rc/wish.rc: removed the #define RESOURCE_INCLUDED lines.
|
|
|
|
| |
wish.exe.manifest to DISTDIR target directory.
|
| |
|
|
|
|
|
|
|
|
|
|
| |
* generic/tkInt.h (struct TkCaret):
* mac/tkMacXStubs.c (Tk_SetCaretPos):
* unix/tkUnixKey.c (TkpGetString, Tk_SetCaretPos):
* win/tkWinX.c (Tk_SetCaretPos):
* tests/tk.test: Added 'tk caret' implementation of TIP#96
* doc/SetCaret.3 (new): which adds a TkCaret structure element to
* doc/tk.n: TkDisplay for maintaining state.
|