ArrowB.c Current rev 1.18 Last Release 1.17
ArrowB.c 1.18
   date: 1998/03/24 23:00:45;  author: rwscott;  state: Exp;  lines: +4 -4
   From:        Eric Howe <mu@wintermute.echo-on.net>
   To:          lesstif@Hungry.COM
   Subject:     Colormap patch
   Date:        Tue, 24 Mar 1998 03:32:40 -0500 (EST)
   
   
   I think I've fixed up LessTif to do the "right thing" when an application
   is using non-default colormaps.  I've only tested the patch with mgv
   (run mgv with the -install switch to get a private colormap) but there
   shouldn't be any problems with other Motif applications.  The patch was
   made against lesstif-current from 1998.03.23.
   
   A summary of the changes follows:
   
   	include/Motif-1.2/XmI/MacrosI.h
   		Added the ColormapOfObject macro.
   
   	include/Motif-1.2/XmI/XmI.h
   		Added prototypes for the new _XmWhitePixelOfObject and
   		_XmBlackPixelOfObject functions (the implementations
   		are in Visual.c).
   
   	lib/Xm/ArrowB.c	lib/Xm/ArrowBG.c
   	lib/Xm/Label.c	lib/Xm/LabelG.c
   	lib/Xm/List.c
   		WhitePixelOfScreen and BlackPixelOfScreen changed to
   		_XmWhitePixelOfObject and _XmBlackPixelOfObject.
   
   	lib/Xm/MainW.c
   	lib/Xm/ScrollBar.c
   	lib/Xm/ScrolledW.c
   		WhitePixelOfScreen and BlackPixelOfScreen changed to
   		_XmWhitePixelOfObject and _XmBlackPixelOfObject;
   		DefaultColormapOfScreen changed to CoreColormap.
   
   	lib/Xm/Visual.c
   		WhitePixelOfScreen and BlackPixelOfScreen changed to
   		_XmWhitePixelOfObject and _XmBlackPixelOfObject;
   		DefaultColormapOfScreen changed to CoreColormap.
   		Added the _XmWhitePixel, _XmBlackPixel,
   		_XmWhitePixelOfObject, and _XmBlackPixelOfObject
   		functions.  _XmWhitePixel and _XmBlackPixel had
   		prototypes in Xm/XmP.h but no implementations,
   		_XmWhitePixelOfObject and _XmBlackPixelOfObject
   		are more convenient forms of _XmWhitePixel and
   		_XmBlackPixel.
   
   Since the drag-and-drop code draws on the root window (AFAIK), I left it
   using the default colormap.
   
   A (very) short summary of what to do when you need a Colormap or black/white
   Pixel value is at
   
   	http://www.echo-on.net/~mu/lesstif-colors.html


ArrowBG.c Current rev 1.13 Last Release 1.12
ArrowBG.c 1.13
   date: 1998/03/24 23:00:47;  author: rwscott;  state: Exp;  lines: +4 -4
   From:        Eric Howe <mu@wintermute.echo-on.net>
   To:          lesstif@Hungry.COM
   Subject:     Colormap patch
   Date:        Tue, 24 Mar 1998 03:32:40 -0500 (EST)
   
   
   I think I've fixed up LessTif to do the "right thing" when an application
   is using non-default colormaps.  I've only tested the patch with mgv
   (run mgv with the -install switch to get a private colormap) but there
   shouldn't be any problems with other Motif applications.  The patch was
   made against lesstif-current from 1998.03.23.
   
   A summary of the changes follows:
   
   	include/Motif-1.2/XmI/MacrosI.h
   		Added the ColormapOfObject macro.
   
   	include/Motif-1.2/XmI/XmI.h
   		Added prototypes for the new _XmWhitePixelOfObject and
   		_XmBlackPixelOfObject functions (the implementations
   		are in Visual.c).
   
   	lib/Xm/ArrowB.c	lib/Xm/ArrowBG.c
   	lib/Xm/Label.c	lib/Xm/LabelG.c
   	lib/Xm/List.c
   		WhitePixelOfScreen and BlackPixelOfScreen changed to
   		_XmWhitePixelOfObject and _XmBlackPixelOfObject.
   
   	lib/Xm/MainW.c
   	lib/Xm/ScrollBar.c
   	lib/Xm/ScrolledW.c
   		WhitePixelOfScreen and BlackPixelOfScreen changed to
   		_XmWhitePixelOfObject and _XmBlackPixelOfObject;
   		DefaultColormapOfScreen changed to CoreColormap.
   
   	lib/Xm/Visual.c
   		WhitePixelOfScreen and BlackPixelOfScreen changed to
   		_XmWhitePixelOfObject and _XmBlackPixelOfObject;
   		DefaultColormapOfScreen changed to CoreColormap.
   		Added the _XmWhitePixel, _XmBlackPixel,
   		_XmWhitePixelOfObject, and _XmBlackPixelOfObject
   		functions.  _XmWhitePixel and _XmBlackPixel had
   		prototypes in Xm/XmP.h but no implementations,
   		_XmWhitePixelOfObject and _XmBlackPixelOfObject
   		are more convenient forms of _XmWhitePixel and
   		_XmBlackPixel.
   
   Since the drag-and-drop code draws on the root window (AFAIK), I left it
   using the default colormap.
   
   A (very) short summary of what to do when you need a Colormap or black/white
   Pixel value is at
   
   	http://www.echo-on.net/~mu/lesstif-colors.html


BulletinBoard.c Current rev 1.24 Last Release 1.21
BulletinBoard.c 1.24
   date: 1998/04/14 22:54:44;  author: rwscott;  state: Exp;  lines: +11 -2
   FileSB.c LabelG.c SelectionBox.c XmString.c Xmos.c
   
   Patch from Peter Wainwright <prw@wainpr.demon.co.uk> dealing mainly with
   memory leaks during string handling.
   
   List.c:_XmListSetGeometry() - If the parent is not managed we were not asking
   	for a geo change.  The result was that a change of visibleItems while
   	the parent was not managed was not changing the visible items.  If the
   	parent is not managed Xt should take care of the size request by
   	just granting it.  This was messing up one of my apps.  If someone
   	can point out why it was there I will re-evaluate this.
   
   BulletinBoard.c:handle_change_managed() - If we are not realized, layout with
   	the current width/height, this will honor the specified resources if
   	any.
BulletinBoard.c 1.23
   date: 1998/03/19 23:10:58;  author: rwscott;  state: Exp;  lines: +16 -8
   _XmBB_CreateButtonG/_XmBBCreateLabelG - If the labelString is NULL do not
   	set it into the args of the create.  This way a value can be picked
   	up from the resource database.
   	(Michel Bardiaux <mbardiaux@usrconsult.be>)
BulletinBoard.c 1.22
   date: 1998/03/14 19:42:07;  author: rwscott;  state: Exp;  lines: +47 -19
   _XmBBParentProcess - if the action is activate and we do not have a default
   	button or dynamic default button buck this up to the parent (if it is
   	a manager).
   	if the action is cancel and we do not have a cancel button buck it
   	up to the parent (if it is a manager).
   
   This is the case of a Form, which has none of the buttons, but is a sub-class
   of BulletinBoard and inherits the parent process.


CascadeB.c Current rev 1.82 Last Release 1.80
CascadeB.c 1.82
   date: 1998/05/08 19:41:49;  author: rwscott;  state: Exp;  lines: +10 -2
   expose() - We were calling resize to make sure that the internal values
   	were correct.  This however falls apart if something is sub-classed
   	from here.  So, lets call the resize method in the class structure
   	which will be the resize method for the sub-class, if it is one, or
   	our resize if not.
   
   This get the BmButtons in Mozilla showing the pixmaps and labels correctly.
CascadeB.c 1.81
   date: 1998/03/16 21:52:05;  author: rwscott;  state: Exp;  lines: +13 -21
   I could not get Andrea's varargs patch to go in clean so I tried it by hand.
   
   defined the rcsid's as const
   
   cleaned out some of the unused variables.


CascadeBG.c Current rev 1.75 Last Release 1.74
CascadeBG.c 1.75
   date: 1998/03/16 21:52:09;  author: rwscott;  state: Exp;  lines: +13 -18
   I could not get Andrea's varargs patch to go in clean so I tried it by hand.
   
   defined the rcsid's as const
   
   cleaned out some of the unused variables.


CutPaste.c Current rev 1.6 Last Release 1.4
CutPaste.c 1.6
   date: 1998/05/14 22:58:01;  author: u27113;  state: Exp;  lines: +45 -5
   Makefile.am's : replace @prefix@ by $(prefix) to make this an installation
   	time option rather than a configure time option.
   
   lib/Xm/TextF.c : Try to get XmClipboard API to work here.
   	Grrr. I've been tracking a typo for hours, but now it does
   	at least something more or less right.
   	All this is still #ifdeffed out.
   lib/Xm/CutPaste.c : add XdbDebugs, one NULL test.
CutPaste.c 1.5
   date: 1998/05/13 21:06:31;  author: u27113;  state: Exp;  lines: +11 -4
   lib/Xm/PanedW.c, lib/Xm-2.0/PanedW.c
   	Jon A. Christopher's fix.
   
   lib/Xm/TextF.c
   	Some work for using XmClipboard API (incomplete)
   
   lib/Xm/CutPaste.c
   	A couple of XdbDebugs + one core dump prevention
   
   CURRENT_NOTES
   	Comment added about using XmClipboard API in XmTextField.


DebugUtil.c Current rev 1.26 Last Release 1.25
DebugUtil.c 1.26
   date: 1998/04/08 21:07:28;  author: rwscott;  state: Exp;  lines: +22 -2
   Add a few more resources for FileSB.c arg lists


DefaultProcs.c Current rev 1.9 Last Release 1.5
DefaultProcs.c 1.9
   date: 1997/08/25 17:40:19;  author: miers;  state: dead;  lines: +2 -2
   Woohoo, we get to get rid of a file.
   
   Two things in this patch.
   
   1) Get Thot's scrollbars to match the background like they should.
   
   2) Move all widget default procs from DefaultProcs.c into the widgets
   themselves.  This allows us to get rid of DefaultProcs.c.
   
   Updated the relevant makefiles to reflect this.
DefaultProcs.c 1.8
   date: 1997/08/18 18:49:27;  author: miers;  state: Exp;  lines: +3 -4
   DefaultProcs: Snuck a traversalOn test in ScrollBar defaults that didn't
   work.  Scales still present a traversal problem.
   
   MessageB.c: Add code into initialize for TEMPLATE checking.
   
   PushB.c: Add NULL event checking from JAC.
   
   Visual.c: Add 1 deep caching for colors from JAC.
   
   XmString.c: Change DrawUnderline to use current_y + line_descent rather
   than current_y + 1 for the underline.  This looks more like Motif.
DefaultProcs.c 1.7
   date: 1997/08/15 23:04:45;  author: miers;  state: Exp;  lines: +4 -3
   Remove extra FIXME's from the CoreClassExtRec(s).  This reduces the list
   to a manageable size (401:)
DefaultProcs.c 1.6
   date: 1997/07/10 01:59:35;  author: miers;  state: Exp;  lines: +100 -42
   Indent fun.
   
   Oh, there's one other thing.  While I'm doing this, I've added a macro
   DEBUGOUT(x) into the DebugUtil.h file.  This macro is empty for
   LESSTIF_PRODUCTION, or expands to its arg when not LESSTIF_PRODUCTION.
   I'm using it as follows:
   
   	DEBUGOUT(XdbDebug(__FILE__, w, "foo was here."));
   
   So that when LESSTIF_PRODUCTION, no debug statement is there.  Danny,
   you may want to change this (maybe another preprocessor flag in
   DebugUtil.h?).
DefaultProcs.c 1.4
   date: 1997/06/04 02:51:06;  author: miers;  state: Exp;  lines: +2 -44
   DefaultProcs.c: Remove dead calls.
   
   GeoUtils.c: Slight rework of _XmGeoLoadValues.
   
   RCUtils.c, RowColumn.c: Real close to there.  I've identified the problem
   areas, but I can't seem to find the combination that gets xmcd and
   xephem both to work.
DefaultProcs.c 1.3
   date: 1997/04/14 06:06:18;  author: u27113;  state: Exp;  lines: +14 -3
   ScrolledW.c : *finally* got the damned thing figured out, after realizing
   	that the scrollbars were misbehaving, which kept me puzzled for weeks.
   
   DebugUtil.c : add function XdbMenuType2String
   DefaultProcs.c : change default for traversing scrollbars in a
   	scrolledwindow. (Traverse when XmAUTOMATIC).
   Primitive.c : add XdbDebug.
   ScrollBar.c : remove test from scrollbar_unhighlight and scrollbar_highlight.
   	Should do Scale stuff via initialisation.
   Traversal.c : add XdbDebug.
DefaultProcs.c 1.2
   date: 1997/03/16 14:17:47;  author: helden;  state: Exp;  lines: +3 -3
   ScrollBar.c:
   	Make it picky again about initial values.
   
   DefaultProcs.c:
   	Motif docs says scrollbar is traversable unless
   	parent is a ScrolledWindow.
   
   RowColumn.c:
   	First shot at implementing menuProcedures. (Not yet used :)
DefaultProcs.c 1.1
   date: 1997/02/20 02:48:09;  author: miers;  state: Exp;
   It's as bad as it looks;)
   
   I HIGHLY recommend to core people that you blow away your source tree, and
   checkout with the prune (-P) option to get rid of empty directories, until
   or unless the empty directories are removed from the repository.
   =============================================================================
   
   RCS file: /cvsroot/hungry/lesstif/lib/Xm/Desktop.c,v
   Working file: Desktop.c
   head: 1.4
   branch:
   locks: strict
   access list:
   symbolic names:
   	release_0_83: 1.4
   	release_0_82: 1.3
   	release_0_81: 1.3
   	release_0_80public: 1.1
   	release_0_80: 1.1
   	release_0_79: 1.1
   	release_0_78: 1.1
   keyword substitution: kv
   total revisions: 4;	selected revisions: 1
   description:
DefaultProcs.c 1.4
   date: 1998/02/22 01:00:27;  author: rwscott;  state: Exp;  lines: +13 -5
   Mostly class structure synchronization, get rid of un-used methods, stub out
   methods that we are supposed to have, get the compress flags right....
   
   A couple of fixes to note.  BulletinBoard was not un-drawing its shadows
   when resized bigger, and the ModifyVerify patch from a few days ago was
   segfaulting on textf/test4.
   
   Now back to the menu system......
   =============================================================================
   
   RCS file: /cvsroot/hungry/lesstif/lib/Xm/Dest.c,v
   Working file: Dest.c
   head: 1.1
   branch:
   locks: strict
   access list:
   symbolic names:
   	release_0_83: 1.1
   	release_0_82: 1.1
   	release_0_81: 1.1
   	release_0_80public: 1.1
   	release_0_80: 1.1
   	release_0_79: 1.1
   	release_0_78: 1.1
   keyword substitution: kv
   total revisions: 1;	selected revisions: 1
   description:
DefaultProcs.c 1.1
   date: 1997/02/20 02:48:11;  author: miers;  state: Exp;
   It's as bad as it looks;)
   
   I HIGHLY recommend to core people that you blow away your source tree, and
   checkout with the prune (-P) option to get rid of empty directories, until
   or unless the empty directories are removed from the repository.
   =============================================================================
   
   RCS file: /cvsroot/hungry/lesstif/lib/Xm/DialogS.c,v
   Working file: DialogS.c
   head: 1.9
   branch:
   locks: strict
   access list:
   symbolic names:
   	release_0_83: 1.9
   	release_0_82: 1.9
   	release_0_81: 1.7
   	release_0_80public: 1.3
   	release_0_80: 1.3
   	release_0_79: 1.1
   	release_0_78: 1.1
   keyword substitution: kv
   total revisions: 9;	selected revisions: 1
   description:
DefaultProcs.c 1.9
   date: 1997/11/07 01:51:38;  author: miers;  state: Exp;  lines: +7 -7
   Change the InitialPlacement to be dependent on whether or not the DialogS
   has an initial position of 0,0.  This still probably isn't right.
   =============================================================================
   
   RCS file: /cvsroot/hungry/lesstif/lib/Xm/Display.c,v
   Working file: Display.c
   head: 1.11
   branch:
   locks: strict
   access list:
   symbolic names:
   	release_0_83: 1.11
   	release_0_82: 1.9
   	release_0_81: 1.8
   	release_0_80public: 1.6
   	release_0_80: 1.6
   	release_0_79: 1.6
   	release_0_78: 1.6
   keyword substitution: kv
   total revisions: 11;	selected revisions: 1
   description:
DefaultProcs.c 1.11
   date: 1998/02/22 01:00:28;  author: rwscott;  state: Exp;  lines: +13 -9
   Mostly class structure synchronization, get rid of un-used methods, stub out
   methods that we are supposed to have, get the compress flags right....
   
   A couple of fixes to note.  BulletinBoard was not un-drawing its shadows
   when resized bigger, and the ModifyVerify patch from a few days ago was
   segfaulting on textf/test4.
   
   Now back to the menu system......
   =============================================================================
   
   RCS file: /cvsroot/hungry/lesstif/lib/Xm/DragBS.c,v
   Working file: DragBS.c
   head: 1.7
   branch:
   locks: strict
   access list:
   symbolic names:
   	release_0_83: 1.7
   	release_0_82: 1.7
   	release_0_81: 1.7
   	release_0_80public: 1.5
   	release_0_80: 1.5
   	release_0_79: 1.5
   	release_0_78: 1.5
   keyword substitution: kv
   total revisions: 7;	selected revisions: 1
   description:
DefaultProcs.c 1.7
   date: 1997/08/25 23:34:29;  author: miers;  state: Exp;  lines: +4 -4
   Some fixes for Insure++ warnings.
   
   Also, the mad indenter hit ScrollBar.  PvH, if you're still working on
   this, just use the last version and overwrite when you're done.
   =============================================================================
   
   RCS file: /cvsroot/hungry/lesstif/lib/Xm/DragC.c,v
   Working file: DragC.c
   head: 1.18
   branch:
   locks: strict
   access list:
   symbolic names:
   	release_0_83: 1.18
   	release_0_82: 1.17
   	release_0_81: 1.16
   	release_0_80public: 1.13
   	release_0_80: 1.13
   	release_0_79: 1.13
   	release_0_78: 1.13
   keyword substitution: kv
   total revisions: 18;	selected revisions: 1
   description:
DefaultProcs.c 1.18
   date: 1998/02/22 01:00:29;  author: rwscott;  state: Exp;  lines: +12 -8
   Mostly class structure synchronization, get rid of un-used methods, stub out
   methods that we are supposed to have, get the compress flags right....
   
   A couple of fixes to note.  BulletinBoard was not un-drawing its shadows
   when resized bigger, and the ModifyVerify patch from a few days ago was
   segfaulting on textf/test4.
   
   Now back to the menu system......
   =============================================================================
   
   RCS file: /cvsroot/hungry/lesstif/lib/Xm/DragDrop.c,v
   Working file: DragDrop.c
   head: 1.5
   branch:
   locks: strict
   access list:
   symbolic names:
   	release_0_83: 1.5
   	release_0_82: 1.5
   	release_0_81: 1.5
   	release_0_80public: 1.2
   	release_0_80: 1.2
   	release_0_79: 1.2
   	release_0_78: 1.2
   keyword substitution: kv
   total revisions: 5;	selected revisions: 1
   description:


DrawingArea.c Current rev 1.21 Last Release 1.20
DrawingArea.c 1.21
   date: 1998/05/04 22:04:36;  author: rwscott;  state: Exp;  lines: +5 -2
   change_managed() - move the enforce margins call inside the if such that
   	it is only called is resizing is allowed.  This gets Kevin's
   	dialogs showing up correctly in the JDK.


FileSB.c Current rev 1.38 Last Release 1.24
FileSB.c 1.38
   date: 1998/04/14 22:54:45;  author: rwscott;  state: Exp;  lines: +9 -5
   FileSB.c LabelG.c SelectionBox.c XmString.c Xmos.c
   
   Patch from Peter Wainwright <prw@wainpr.demon.co.uk> dealing mainly with
   memory leaks during string handling.
   
   List.c:_XmListSetGeometry() - If the parent is not managed we were not asking
   	for a geo change.  The result was that a change of visibleItems while
   	the parent was not managed was not changing the visible items.  If the
   	parent is not managed Xt should take care of the size request by
   	just granting it.  This was messing up one of my apps.  If someone
   	can point out why it was there I will re-evaluate this.
   
   BulletinBoard.c:handle_change_managed() - If we are not realized, layout with
   	the current width/height, this will honor the specified resources if
   	any.
FileSB.c 1.37
   date: 1998/04/10 14:05:49;  author: rwscott;  state: Exp;  lines: +18 -6
   XmFileSelectionDoSearch() - Make xmgr happy.  If filter text is not managed
   	do not use it, use directory instead.
   
   This is still not done.  I need some more monitoring of the FSB to find out
   what is changing and when.
FileSB.c 1.36
   date: 1998/04/08 12:21:56;  author: rwscott;  state: Exp;  lines: +46 -2
   Ooopps, I missed XmFileSelectionDoSearch yesterday.  It is almost the
   same as a filter operation with the exception that the dirmask may be
   specified.
FileSB.c 1.35
   date: 1998/04/08 00:12:09;  author: rwscott;  state: Exp;  lines: +210 -54
   FileSB.c:defaultQualifySearchDataProc() - Completely re-write this puppy.
   	Use filesb/test9 to check it out.
   FileSB.c:_XmFsbButton() - Use the Qualify proc to do the filtering.
   
   Xmos.c:_XmOSQualifyFileSpec() - Never return a relative path directory. If
   	it comes in relative tack on the cwd.
   
   Beware, I still have some questions about this.  I have not tried it with
   Mozilla yet but, it seems to keep xmgr happy even with the strange things
   it is doing.
FileSB.c 1.34
   date: 1998/04/06 19:44:34;  author: rwscott;  state: Exp;  lines: +49 -17
   Start to take a close look at how this puppy works.  No operational changes
   in this I just need to check it in to pick it up at another location.
FileSB.c 1.33
   date: 1998/04/04 03:07:05;  author: rwscott;  state: Exp;  lines: +34 -7
   Related to segfaults and infinite loops with file selection in Mozilla.
   This gets the dialog to show, but traversing the file system is another thing.
   It uses its own qualify/dir/file search procs. That do not seem to be all
   that compatible with how we do things.  Now to locate them in the Mozilla
   source to see what is going on!
FileSB.c 1.32
   date: 1998/03/25 21:57:11;  author: rwscott;  state: Exp;  lines: +18 -4
   XmFileSelectionDoSearch() - Apparently the fix for xmgr was not totally
   	correct, since it messed up mgv, Mosaic and probably others. So now
   	we set the DirMask from the filter text if it is managed, otherwise
   	set the Directory from the SB_Text, unless it is not managed.
   	I hate having 3 resources to set 2 values!
FileSB.c 1.31
   date: 1998/03/19 22:24:30;  author: rwscott;  state: Exp;  lines: +7 -2
   List.c:_XmListSetGeometry() - Only make a geometry request to the parent if
   	said parent is managed. (xmgr working directory)
   
   FileSB.c:XmFileSelectionDoSearch() - Apparently if the dirMask supplied to
   	this routine is NULL, the dirMask and directory are set from the
   	corresponding textFields.  Up until now we where only extracting
   	the dirMask. (xmgr work directory)
FileSB.c 1.30
   date: 1998/03/19 20:55:01;  author: rwscott;  state: Exp;  lines: +25 -4
   FileSB.c:focus_change_proc() - use this to change the BB dynamic default
   	button.  Now that parent_process is passing events up pressing
   	return in the filter text field causes the OK button to be pressed.
   	by changing the dynamic default button depending on where the focus
   	is fixes this.
   
   A little protection to my earlier patch to provide a minimum width for the
   dir list for when most of the sub-widgets are un-managed as in xmgr.
   As far as xmgr goes, its working directory dialog is un-managing most of
   the widgets in a FileSB, changing the selection mode of the dirList, and
   providing its own callbacks.  Is it any wonder it is having problems!
FileSB.c 1.29
   date: 1998/03/19 20:07:03;  author: u27113;  state: Exp;  lines: +3 -3
   FileSB.c: something in Eric Howe's patch drove xmgr into an
   	infinite loop. Fixed that.
   TrackLoc.c : Andreas Zeller's patch.
FileSB.c 1.28
   date: 1998/03/19 17:07:27;  author: rwscott;  state: Exp;  lines: +3 -2
   FileSB.c:_XmDirListFix() - set the preferred height of the list to its
   	current height.
   
   MenuShell.c:MenuShellPopupSharedMenuPane() - We were trying to resize an
   	object to 0x0.  This showed in the nedit open previous file menu.
FileSB.c 1.27
   date: 1998/03/19 16:16:38;  author: rwscott;  state: Exp;  lines: +33 -4
   GeoUtils.c:_XmHandleSizeUpdate() - If the policy was RESIZE_GROW and the
   	width wanted to be larger but the height smaller both changes were
   	rejected.  Now reject only the one that is smaller than the current.
   	(Playing in the GeoUtils, does this bring back bad memories!)
   
   FileSB.c:noGeoRequest() - Always return True here because we really do not
   	care how big the lists want to be, we set them in UpdateSize. This
   	may be a little restrictive, what if one of the texts want to change
   	size????
   
   Lots more debug stuff.
   
   This gets the nedit and filesb/test8 file selection correct.
FileSB.c 1.26
   date: 1998/03/19 12:34:14;  author: rwscott;  state: Exp;  lines: +117 -31
   patch from Eric Howe <mu@wintermute.echo-on.net>
   A patch which fixes the following problems is at the end of this message:
   
   lib/Xm/FileSB.c
   	The file list is now sorted just like it is in Motif.
   	The selected item in the directory list is now reset to the
   	first item when you change directories (Motif behaves this way).
   	The fake " [ .. ] " placeholder in an empty file list is no longer
   	selectable.  The second fix uses the XmFS_NO_MATCH,
   	XmFS_IN_FILE_SEARCH, and XmFS_DIR_SEARCH_PROC flags and the
   	XmFileSelectionBoxPart.state_flags field to avoid some infinite loops,
   	I think I'm using these flags as intended (they were not used before
   	this).
   
   	These fixes required a couple of List_LastHLItem(w) fixes to
   	lib/Xm/List.c to avoid assertion failures and core dumps.
   
   lib/Xm/List.c
   	Changing pages with Page|Next or Page|Previous in MGv would leave
   	multiple list items visually selected (i.e. in reverse video).
   	_XmListDeselectPos was not being called in XmListSelectPos when the
   	selectionPolicy was single- or browse-select (hence, the old item
   	would remain visually selected); XmListSelectPos also calls
   	_XmListSetCursorPos now to get the highlight right.  This problem
   	should have appeared in any application that called XmListSelectPos.
FileSB.c 1.25
   date: 1998/03/18 23:51:59;  author: rwscott;  state: Exp;  lines: +10 -2
   Do not let either of the lists to be narrower than its corresponding
   label.


FontList.c Current rev 1.13 Last Release 1.10
FontList.c 1.13
   date: 1998/04/16 19:19:42;  author: u27113;  state: Exp;  lines: +4 -3
   Text.c : back out change to XmTextSetString. This should fix complaints
   	about it changing the position shown when it shouldn't.
   
   Others: as indicated by Alexander Mai <mai@migdal.ikp.physik.tu-darmstadt.de>,
   	#include <LTconfig.h> should precede (almost) all other includes
   	in the sources.
FontList.c 1.12
   date: 1998/03/25 21:09:48;  author: u27113;  state: Exp;  lines: +11 -8
   FontList.c : remove some memory leaks;
   	more importantly: don't return our argument when NULL passed
   	in XmFontListAppendEntry; this doesn't work with the memory
   	allocation rules.
   
   TextF.c (both) : watch out more carefully about when you XmFontListFree.
   	Yesterday's implementation did one where it shouldn't.
   	Also do whatever's necessary when changing the XmNfontList
   	through SetValues. Now the newly adapted test/textf/test2.c works.
   
   test/textf/test2.c : call setvalues on XmNfontList to show what didn't work.
   
   ScrolledW.c: add some XdbDebugs, also strip a couple of bits off what
   	GeometryManager returns.
FontList.c 1.11
   date: 1998/03/18 19:43:12;  author: u27113;  state: Exp;  lines: +3 -3
   lib/Xm/FontList.c : correction by Kevin B. Hendricks for a fix for the
   	JDK port to Linux. <kbhend@dogwood.tyler.wm.edu>
   
   lib/Xm/SelectionBox.c : fix by Andreas Zeller <zeller@acm.org>


Form.c Current rev 1.109 Last Release 1.106
Form.c 1.109
   date: 1998/05/09 19:57:40;  author: rwscott;  state: Exp;  lines: +10 -10
   _XmFormConfigureChildren() - If a child was to be configured for 0x0 we would
   	bump this up to 1x1 to avoid problems later on. Turns out this is
   	causing more problems than it helps.  If a child was being added
   	to a _managed_ form then this size would stick.  So let's allow 0x0
   	now because I think that we provide protection bdfore most of the
   	window related things now.
   
   This was causing the missing menu bar problem, recently reported, and probably
   the one with the JDK.  Also responsible for the squashed scrolled window in
   Xquote->stock symbols and probably a few more.  Basically it would cause
   problems whenever a sub-class of Manager was added to a _managed_ form.
   
   PS
   	Managers should never be created managed.  They should be created
   un-managed and only managed when all of the children have been added to them.
   This saves a round of geometry negotiation for each child that is added.  It
   is _much_ better to add all of the children to the Manager and then manage it.
   This way only one round of geo. negotiation takes place.
Form.c 1.108
   date: 1998/04/15 21:57:06;  author: rwscott;  state: Exp;  lines: +32 -2
   Change the constraint synthetic resource import/export procs for the offsets.
   Offsets are not resolution independant (although I cannot find where it says
   that right now), so just check for INVALID_DIMENSION and if it is return 0.
   I thought that it should be vertical/horizontal Spacing but Motif does not
   seem to do this (form/test34).
Form.c 1.107
   date: 1998/03/25 02:59:58;  author: rwscott;  state: Exp;  lines: +14 -2
   geometry_manager() - If the form requests a new size from its parent and the
   	parent says NO, then run the layout again with the requested geom
   	and the current size of the form.  This re-distributes the kids
   	within the current size.
   
   This is for the mgv magstep problem.  This is a pretty fundamental change to
   Form's geometry management, let me know as soon as possible if it screws up
   anything else please.


Frame.c Current rev 1.35 Last Release 1.34
Frame.c 1.35
   date: 1998/04/03 14:22:46;  author: rwscott;  state: Exp;  lines: +11 -2
   geometry_manager() - If the work area child asks for a size of 0x0
   	print a warning and refuse it!  This may be a problem in SW instead.
   	Should it be asking for a size of 0x0??
   
   Mozilla X Errors to configure window on startup


GadgetUtil.c Current rev 1.7 Last Release 1.6
GadgetUtil.c 1.7
   date: 1998/04/09 12:23:44;  author: u27113;  state: Exp;  lines: +10 -2
   lib/Xm/GadgetUtil.c : Rename _XmInputInGadget for 2.0 for Mozilla.
   lib/Xm/ScrolledW.c : Don't ask for new geometry if not realized.
   	Maybe this fixes Alexander Mai's XACC problem.
   lib/Xm/TextF.c : Clean up some junk.
   lib/Xm/misc.c : Mozilla build script looks for an identifying string
   	in the shared library (in xmversion.sh).
   doc/lessdox/Makefile.am : add subdirectory widgets


GeoUtils.c Current rev 1.22 Last Release 1.18
GeoUtils.c 1.22
   date: 1998/04/10 12:35:37;  author: u27113;  state: Exp;  lines: +10 -4
   GeoUtils.c : Yet another fix to avoid Xt problems.
   	This will hopefully avoid these nasty X_ConfigureWindow
   	X Errors. Apparently having a 0 height in a widget before
   	calling XtMakeGeometryRequest can kill us too.
   test : extend to be able to show one of Andreas Zeller's bug reports
   	Apparently this one was fixed already: Items in a selectionbox
   	can be overruled by setting XmNlistLabelString.
GeoUtils.c 1.21
   date: 1998/04/10 09:23:55;  author: u27113;  state: Exp;  lines: +19 -2
   doc/lessdox/widgets/XmFileSelectionBox.3
   	A small amount of real documentation.
   lib/Xm/ScrolledW.c
   	Remove yesterday's hack.
   lib/Xm/GeoUtils.c
   	To put it here :-)
   
   	In _XmMakeGeometryRequest, if we will be requesting a 0 width
   	or height, change that to 1 to prevent Xt from doing strange
   	things (like X errors).
   
   	This fixes Alexander Mai's xacc problem (message dated 6 April).
   	It also fixes the scrolling lists in selection boxes that
   	behaved rather strange lately.
   	Doesn't fix Xinvest yet.
GeoUtils.c 1.20
   date: 1998/03/29 21:57:05;  author: rwscott;  state: Exp;  lines: +7 -5
   GeoUtils.c:_XmHandleGeometryManager() - If the reisize policy permits the
   	request we check to make sure that the instigator got its requested
   	size. If it did we request the new size from the parent. However
   	we had a ! in the test so that we would request a size change from
   	the parent only if the instigator did not get the size it wanted. So
   	we take out the !.  If the instigator did not get the size it wanted
   	return Almost.  This gives a second chance in case only 1 of the
   	dimensions was not as requested.  For example if a menu bar changes
   	its height, the new height may be honoured but a MessageBox will set
   	the width to the width of the box.  Returning Almost in this case
   	gives it a second chance.
   
   ScrolledW.c:_XmScrolledWGeomRequest() - Fixing the GeoUtils allows this ugly
   	MFM_BUGS hack to be removed, I think.
GeoUtils.c 1.19
   date: 1998/03/19 16:16:40;  author: rwscott;  state: Exp;  lines: +27 -9
   GeoUtils.c:_XmHandleSizeUpdate() - If the policy was RESIZE_GROW and the
   	width wanted to be larger but the height smaller both changes were
   	rejected.  Now reject only the one that is smaller than the current.
   	(Playing in the GeoUtils, does this bring back bad memories!)
   
   FileSB.c:noGeoRequest() - Always return True here because we really do not
   	care how big the lists want to be, we set them in UpdateSize. This
   	may be a little restrictive, what if one of the texts want to change
   	size????
   
   Lots more debug stuff.
   
   This gets the nedit and filesb/test8 file selection correct.


Hash.c Current rev 1.3 Last Release 1.2
Hash.c 1.3
   date: 1998/04/16 19:19:42;  author: u27113;  state: Exp;  lines: +2 -0
   Text.c : back out change to XmTextSetString. This should fix complaints
   	about it changing the position shown when it shouldn't.
   
   Others: as indicated by Alexander Mai <mai@migdal.ikp.physik.tu-darmstadt.de>,
   	#include <LTconfig.h> should precede (almost) all other includes
   	in the sources.


LTEditres.c Current rev 1.6 Last Release 1.1
LTEditres.c 1.6
   date: 1997/08/19 19:29:08;  author: miers;  state: dead;  lines: +0 -0
   More code reorg.
   
   1) Moved all the Xme functions to the 2.0 tree.  Something needs to be
   done about the headers, though.
   
   2) If LESSTIF_PRODUCTION is defined, then the Editres handler and the
      pixmap extensions aren't built.  I'd like to move these to another
      place (e.g., extensions/) but it's so much trouble to hack configure
      that I'm putting it off.
   
   3) Add an Editres for R5.
LTEditres.c 1.5
   date: 1997/07/26 20:22:09;  author: miers;  state: Exp;  lines: +1 -1
   DialogS.c: We seem to need the XtUnmapWidget (nedit->Open->Ok, without
   selecting anything).  However, move it to after the XtPopdown call.
   
   Form.c: Added a #define, NO_QUERY_GEO (currently #undef'ed) that replaces
   the XtQueryGeometry call with retrieval of w/h/bw.  This mostly works,
   except in the case where the form changes (see test15,
   xephem->Earth View->Spherical); something more is needed before the XtQuery
   can go for good.  Rick, you may want to take a look, you know this puppy
   better than me.  Maybe this could be where InitialWidth/Height comes into
   play?  Dunno.
   
   LTEditres.c, Label.c, LabelG.c, ResConvert.c: Warning removal.
   
   Vendor.c: Cleanup some of the dead code.
LTEditres.c 1.4
   date: 1997/07/11 02:30:01;  author: miers;  state: Exp;  lines: +1 -0
   Alter DebugUtil so it doesn't need any Motif headers for use.
   For LTEditres.
LTEditres.c 1.3
   date: 1997/07/11 02:22:17;  author: miers;  state: Exp;  lines: +985 -887
   Indent.
LTEditres.c 1.2
   date: 1997/07/08 18:41:49;  author: u27113;  state: Exp;  lines: +221 -0
   lib/Xm/LTEditres.c, lib/Xm/ResConvert.c : Changes to add EditRes GET behaviour,
   	including a couple of resource converters (*-to-string), including
   	int-to-string, XmString-to-string. Check out e.g. labelString on any
   	XmLabel subclass.
   lib/Xm/TextOut.c : these fixes should make the thing more stable.
   	Fixed a number of PASTENDPOS bugs.
   test/Xm/form/test6.c : expanded test.


LTV6Editres.c Current rev 1.3 Last Release 1.2
LTV6Editres.c 1.3
   date: 1998/04/01 20:29:28;  author: u27113;  state: Exp;  lines: +0 -18
   Edit out the old editres handlers (before the patches that make
   Editres also handle GETting data).
   
   Also remove an #if; now Editres is in there even when LESSTIF_PRODUCTION.


Label.c Current rev 1.63 Last Release 1.58
Label.c 1.63
   date: 1998/05/08 19:41:50;  author: rwscott;  state: Exp;  lines: +10 -2
   expose() - We were calling resize to make sure that the internal values
   	were correct.  This however falls apart if something is sub-classed
   	from here.  So, lets call the resize method in the class structure
   	which will be the resize method for the sub-class, if it is one, or
   	our resize if not.
   
   This get the BmButtons in Mozilla showing the pixmaps and labels correctly.
Label.c 1.62
   date: 1998/05/04 21:47:49;  author: rwscott;  state: Exp;  lines: +10 -3
   set_values() - A while ago I added a relayout when the sensitivity changes
   	because if the type is a pixmap the pixmap size may change.  This
   	however screwed up the file list of plan.  So wrap the relayout_needed
   	in if Lab_RecomputeSize like all the other areas.
Label.c 1.61
   date: 1998/04/16 23:32:53;  author: rwscott;  state: Exp;  lines: +11 -3
   set_values() - If the sensitivity changes the size may also change, especially
   	if the label is showing a pixmap.
   
   _XmCalcLabel[G]Dimensions() - If the label is a pixmap calculate the size
   	based on whether or not the label is sensitive.
Label.c 1.60
   date: 1998/03/24 23:00:50;  author: rwscott;  state: Exp;  lines: +10 -16
   From:        Eric Howe <mu@wintermute.echo-on.net>
   To:          lesstif@Hungry.COM
   Subject:     Colormap patch
   Date:        Tue, 24 Mar 1998 03:32:40 -0500 (EST)
   
   
   I think I've fixed up LessTif to do the "right thing" when an application
   is using non-default colormaps.  I've only tested the patch with mgv
   (run mgv with the -install switch to get a private colormap) but there
   shouldn't be any problems with other Motif applications.  The patch was
   made against lesstif-current from 1998.03.23.
   
   A summary of the changes follows:
   
   	include/Motif-1.2/XmI/MacrosI.h
   		Added the ColormapOfObject macro.
   
   	include/Motif-1.2/XmI/XmI.h
   		Added prototypes for the new _XmWhitePixelOfObject and
   		_XmBlackPixelOfObject functions (the implementations
   		are in Visual.c).
   
   	lib/Xm/ArrowB.c	lib/Xm/ArrowBG.c
   	lib/Xm/Label.c	lib/Xm/LabelG.c
   	lib/Xm/List.c
   		WhitePixelOfScreen and BlackPixelOfScreen changed to
   		_XmWhitePixelOfObject and _XmBlackPixelOfObject.
   
   	lib/Xm/MainW.c
   	lib/Xm/ScrollBar.c
   	lib/Xm/ScrolledW.c
   		WhitePixelOfScreen and BlackPixelOfScreen changed to
   		_XmWhitePixelOfObject and _XmBlackPixelOfObject;
   		DefaultColormapOfScreen changed to CoreColormap.
   
   	lib/Xm/Visual.c
   		WhitePixelOfScreen and BlackPixelOfScreen changed to
   		_XmWhitePixelOfObject and _XmBlackPixelOfObject;
   		DefaultColormapOfScreen changed to CoreColormap.
   		Added the _XmWhitePixel, _XmBlackPixel,
   		_XmWhitePixelOfObject, and _XmBlackPixelOfObject
   		functions.  _XmWhitePixel and _XmBlackPixel had
   		prototypes in Xm/XmP.h but no implementations,
   		_XmWhitePixelOfObject and _XmBlackPixelOfObject
   		are more convenient forms of _XmWhitePixel and
   		_XmBlackPixel.
   
   Since the drag-and-drop code draws on the root window (AFAIK), I left it
   using the default colormap.
   
   A (very) short summary of what to do when you need a Colormap or black/white
   Pixel value is at
   
   	http://www.echo-on.net/~mu/lesstif-colors.html
Label.c 1.59
   date: 1998/03/15 12:44:26;  author: rwscott;  state: Exp;  lines: +4 -3
   _XmLabelShowsAccelerators() - Patch from Eric Howe <mu@wintermute.echo-on.net>
   	to allow accelerators to show up in popupmenus.


LabelG.c Current rev 1.54 Last Release 1.51
LabelG.c 1.54
   date: 1998/04/16 23:32:54;  author: rwscott;  state: Exp;  lines: +11 -3
   set_values() - If the sensitivity changes the size may also change, especially
   	if the label is showing a pixmap.
   
   _XmCalcLabel[G]Dimensions() - If the label is a pixmap calculate the size
   	based on whether or not the label is sensitive.
LabelG.c 1.53
   date: 1998/04/14 22:54:46;  author: rwscott;  state: Exp;  lines: +10 -5
   FileSB.c LabelG.c SelectionBox.c XmString.c Xmos.c
   
   Patch from Peter Wainwright <prw@wainpr.demon.co.uk> dealing mainly with
   memory leaks during string handling.
   
   List.c:_XmListSetGeometry() - If the parent is not managed we were not asking
   	for a geo change.  The result was that a change of visibleItems while
   	the parent was not managed was not changing the visible items.  If the
   	parent is not managed Xt should take care of the size request by
   	just granting it.  This was messing up one of my apps.  If someone
   	can point out why it was there I will re-evaluate this.
   
   BulletinBoard.c:handle_change_managed() - If we are not realized, layout with
   	the current width/height, this will honor the specified resources if
   	any.
LabelG.c 1.52
   date: 1998/03/24 23:00:53;  author: rwscott;  state: Exp;  lines: +10 -16
   From:        Eric Howe <mu@wintermute.echo-on.net>
   To:          lesstif@Hungry.COM
   Subject:     Colormap patch
   Date:        Tue, 24 Mar 1998 03:32:40 -0500 (EST)
   
   
   I think I've fixed up LessTif to do the "right thing" when an application
   is using non-default colormaps.  I've only tested the patch with mgv
   (run mgv with the -install switch to get a private colormap) but there
   shouldn't be any problems with other Motif applications.  The patch was
   made against lesstif-current from 1998.03.23.
   
   A summary of the changes follows:
   
   	include/Motif-1.2/XmI/MacrosI.h
   		Added the ColormapOfObject macro.
   
   	include/Motif-1.2/XmI/XmI.h
   		Added prototypes for the new _XmWhitePixelOfObject and
   		_XmBlackPixelOfObject functions (the implementations
   		are in Visual.c).
   
   	lib/Xm/ArrowB.c	lib/Xm/ArrowBG.c
   	lib/Xm/Label.c	lib/Xm/LabelG.c
   	lib/Xm/List.c
   		WhitePixelOfScreen and BlackPixelOfScreen changed to
   		_XmWhitePixelOfObject and _XmBlackPixelOfObject.
   
   	lib/Xm/MainW.c
   	lib/Xm/ScrollBar.c
   	lib/Xm/ScrolledW.c
   		WhitePixelOfScreen and BlackPixelOfScreen changed to
   		_XmWhitePixelOfObject and _XmBlackPixelOfObject;
   		DefaultColormapOfScreen changed to CoreColormap.
   
   	lib/Xm/Visual.c
   		WhitePixelOfScreen and BlackPixelOfScreen changed to
   		_XmWhitePixelOfObject and _XmBlackPixelOfObject;
   		DefaultColormapOfScreen changed to CoreColormap.
   		Added the _XmWhitePixel, _XmBlackPixel,
   		_XmWhitePixelOfObject, and _XmBlackPixelOfObject
   		functions.  _XmWhitePixel and _XmBlackPixel had
   		prototypes in Xm/XmP.h but no implementations,
   		_XmWhitePixelOfObject and _XmBlackPixelOfObject
   		are more convenient forms of _XmWhitePixel and
   		_XmBlackPixel.
   
   Since the drag-and-drop code draws on the root window (AFAIK), I left it
   using the default colormap.
   
   A (very) short summary of what to do when you need a Colormap or black/white
   Pixel value is at
   
   	http://www.echo-on.net/~mu/lesstif-colors.html


List.c Current rev 1.50 Last Release 1.38
List.c 1.50
   date: 1998/05/08 12:53:25;  author: rwscott;  state: Exp;  lines: +10 -8
   misc/test16 shows that visibleItemCount starts life at 0. This allows us
   to catch it in initialize and set it to numItems if it has not been specified.
   This makes scrolledwindow/test14 display correctly.
   
   Xquote Options->ticker symbols is still messed up.  Somewhere the scrolled
   window is getting resized with a height of 1, exactly what is showing.
   However the list has visibleItems set to 5 so this should not be happening!
List.c 1.49
   date: 1998/04/14 22:54:47;  author: rwscott;  state: Exp;  lines: +9 -2
   FileSB.c LabelG.c SelectionBox.c XmString.c Xmos.c
   
   Patch from Peter Wainwright <prw@wainpr.demon.co.uk> dealing mainly with
   memory leaks during string handling.
   
   List.c:_XmListSetGeometry() - If the parent is not managed we were not asking
   	for a geo change.  The result was that a change of visibleItems while
   	the parent was not managed was not changing the visible items.  If the
   	parent is not managed Xt should take care of the size request by
   	just granting it.  This was messing up one of my apps.  If someone
   	can point out why it was there I will re-evaluate this.
   
   BulletinBoard.c:handle_change_managed() - If we are not realized, layout with
   	the current width/height, this will honor the specified resources if
   	any.
List.c 1.48
   date: 1998/04/10 19:17:17;  author: jonf;  state: Exp;  lines: +121 -43
   1.	Added missing lines to _XmListRecalcItemSize().
   	I forgot to add the important lines that set
   	List_MaxWidth and List_MaxItemHeight.  Oops.
   
   2.	Disabled a block of code that messed with the
   	scrollbars in _XmListSetGeometry(), because it
   	would sometimes screw up the scrollbars during
   	geometry negotiations.
   
   3.	Fixed a minor problem that would cause list items to be
   	off by one pixel in width.
   
   4.	Changed the max value of the horizontal scrollbar to match
   	the way Motif defines.  Motif defines the scrolling area as
   	(max width of all items) + 2 * (shadow + highlight + 1)
   
   These changes fix bugs in the following programs:
   
   Xinvest(list size shrinks when XmListAddItemUnselected() called)
   
   ddd(clicking on list doesn't clear items already selected)
   
   MGv(page list size problems)
List.c 1.47
   date: 1998/03/31 02:25:44;  author: jonf;  state: Exp;  lines: +121 -58
   Fixed keyboard up/down actions in scrolling list.  Original code
   would not scroll beyond the visible items.  I thought this had
   been working before...
   
   Fixed problem with XmVARIABLE implementation.  I hope this
   is right.  The old code wouldn't resize the width if the max width
   of the items was reduced.  This caused problems in mgv.  The
   new code will resizes the width as needed, and recalculates
   List_MaxWidth(w) and List_MaxItemHeight(w), so that the geometry
   calculation correctly resizes the list.
   
   Fixed problem with setting XmNselectedItems at create time.  This
   was not previously implemented.
   
   Fixed problem when changing selection mode from XmMULTIPLE_SELECT
   or XmEXTENDED_SELECT to XmSINGLE_SELECT or XmBROWSE_SELECT.  It
   now clears the multiple selections before selecting a new item.
   
   Renamed _XmListRecalcItems() to _XmListCalcVisibleItemCount()
   to make the name more meaningful.
   
   Created new static function _XmListRecalcItemSize(), which
   updates List_MaxWidth(w) and List_MaxItemHeight(w).
List.c 1.46
   date: 1998/03/28 01:42:01;  author: rwscott;  state: Exp;  lines: +4 -4
   ListBestHeight() - After adding a couple of hundred items it became clear
   	that the fudge for the height I added earlier should not be related
   	to the visibleItemCount but a constant.
List.c 1.45
   date: 1998/03/27 22:08:16;  author: rwscott;  state: Exp;  lines: +3 -6
   _XmListSetGeometry() - Make the geometry request to the parent even if the
   	parent is not managed.  This call was taken out for the xmgr working
   	dialog.  Subsequent work on FSB and List allow this call to go back
   	in.  Besides, one of my apps needs this call and, if it comes down to
   	xmgr vs. one of mine, mine will win every time :)
List.c 1.44
   date: 1998/03/27 13:38:17;  author: rwscott;  state: Exp;  lines: +18 -5
   _XmListBestWidth() -
          This special case is returning essentially the ListBestHeight for the
          ListBestWidth, which seems a little wacky.  Let's make sure the
          special case does not get triggered and see what happens.  This shows
          up in mgv Page->Show Page Labels.
   
   _XmListBestHeight() - list/test6 is supposed to show 5 items and it only shows
   	4.75 so I added visibleItemCount - 1 pixels to the height.
   
   set_values() - If new items are set here do not reset the top_position to
   	1. list/test6 sets both in the same call and because of this the
   	topPosition was being ignored.
List.c 1.43
   date: 1998/03/25 22:11:26;  author: rwscott;  state: Exp;  lines: +3 -3
   XmListSetBottomPos() - typo List_itemCount should be List_ItemCount
List.c 1.42
   date: 1998/03/25 19:58:49;  author: u27113;  state: Exp;  lines: +22 -12
   Let W be a List widget with 0 members, and let us call
   XmListSelectPos() with W and a POSITION of 0.  In LessTif 0.83, this
   causes a segmentation fault:
   
   1. XmListSelectPos() is called with POSITION = 0
   2. _XmListSelectPos() is called with POSITION = 0
   3. _XmListSelectPos() finds that 0 means the last element, and
      sets POSITION to 0 (the index of the last element)
   4. _XmListSelectPos() accesses List_InternalList(w)[POSITION - 1],
      which results in a segmentation fault.
   
   Solution: check whether the List actually contains some elements.  A
   patch (which checks for similar situations in other List functions) is
   attached.
   
   Happy hacking,
   
   --
   Andreas Zeller     Technische Universitaet Braunschweig, Germany
                      mailto:zeller@acm.org http://www.cs.tu-bs.de/~zeller/
List.c 1.41
   date: 1998/03/24 23:01:00;  author: rwscott;  state: Exp;  lines: +4 -4
   From:        Eric Howe <mu@wintermute.echo-on.net>
   To:          lesstif@Hungry.COM
   Subject:     Colormap patch
   Date:        Tue, 24 Mar 1998 03:32:40 -0500 (EST)
   
   
   I think I've fixed up LessTif to do the "right thing" when an application
   is using non-default colormaps.  I've only tested the patch with mgv
   (run mgv with the -install switch to get a private colormap) but there
   shouldn't be any problems with other Motif applications.  The patch was
   made against lesstif-current from 1998.03.23.
   
   A summary of the changes follows:
   
   	include/Motif-1.2/XmI/MacrosI.h
   		Added the ColormapOfObject macro.
   
   	include/Motif-1.2/XmI/XmI.h
   		Added prototypes for the new _XmWhitePixelOfObject and
   		_XmBlackPixelOfObject functions (the implementations
   		are in Visual.c).
   
   	lib/Xm/ArrowB.c	lib/Xm/ArrowBG.c
   	lib/Xm/Label.c	lib/Xm/LabelG.c
   	lib/Xm/List.c
   		WhitePixelOfScreen and BlackPixelOfScreen changed to
   		_XmWhitePixelOfObject and _XmBlackPixelOfObject.
   
   	lib/Xm/MainW.c
   	lib/Xm/ScrollBar.c
   	lib/Xm/ScrolledW.c
   		WhitePixelOfScreen and BlackPixelOfScreen changed to
   		_XmWhitePixelOfObject and _XmBlackPixelOfObject;
   		DefaultColormapOfScreen changed to CoreColormap.
   
   	lib/Xm/Visual.c
   		WhitePixelOfScreen and BlackPixelOfScreen changed to
   		_XmWhitePixelOfObject and _XmBlackPixelOfObject;
   		DefaultColormapOfScreen changed to CoreColormap.
   		Added the _XmWhitePixel, _XmBlackPixel,
   		_XmWhitePixelOfObject, and _XmBlackPixelOfObject
   		functions.  _XmWhitePixel and _XmBlackPixel had
   		prototypes in Xm/XmP.h but no implementations,
   		_XmWhitePixelOfObject and _XmBlackPixelOfObject
   		are more convenient forms of _XmWhitePixel and
   		_XmBlackPixel.
   
   Since the drag-and-drop code draws on the root window (AFAIK), I left it
   using the default colormap.
   
   A (very) short summary of what to do when you need a Colormap or black/white
   Pixel value is at
   
   	http://www.echo-on.net/~mu/lesstif-colors.html
List.c 1.40
   date: 1998/03/19 22:24:31;  author: rwscott;  state: Exp;  lines: +6 -3
   List.c:_XmListSetGeometry() - Only make a geometry request to the parent if
   	said parent is managed. (xmgr working directory)
   
   FileSB.c:XmFileSelectionDoSearch() - Apparently if the dirMask supplied to
   	this routine is NULL, the dirMask and directory are set from the
   	corresponding textFields.  Up until now we where only extracting
   	the dirMask. (xmgr work directory)
List.c 1.39
   date: 1998/03/19 12:34:16;  author: rwscott;  state: Exp;  lines: +13 -3
   patch from Eric Howe <mu@wintermute.echo-on.net>
   A patch which fixes the following problems is at the end of this message:
   
   lib/Xm/FileSB.c
   	The file list is now sorted just like it is in Motif.
   	The selected item in the directory list is now reset to the
   	first item when you change directories (Motif behaves this way).
   	The fake " [ .. ] " placeholder in an empty file list is no longer
   	selectable.  The second fix uses the XmFS_NO_MATCH,
   	XmFS_IN_FILE_SEARCH, and XmFS_DIR_SEARCH_PROC flags and the
   	XmFileSelectionBoxPart.state_flags field to avoid some infinite loops,
   	I think I'm using these flags as intended (they were not used before
   	this).
   
   	These fixes required a couple of List_LastHLItem(w) fixes to
   	lib/Xm/List.c to avoid assertion failures and core dumps.
   
   lib/Xm/List.c
   	Changing pages with Page|Next or Page|Previous in MGv would leave
   	multiple list items visually selected (i.e. in reverse video).
   	_XmListDeselectPos was not being called in XmListSelectPos when the
   	selectionPolicy was single- or browse-select (hence, the old item
   	would remain visually selected); XmListSelectPos also calls
   	_XmListSetCursorPos now to get the highlight right.  This problem
   	should have appeared in any application that called XmListSelectPos.


MainW.c Current rev 1.28 Last Release 1.26
MainW.c 1.28
   date: 1998/03/24 23:01:04;  author: rwscott;  state: Exp;  lines: +5 -6
   From:        Eric Howe <mu@wintermute.echo-on.net>
   To:          lesstif@Hungry.COM
   Subject:     Colormap patch
   Date:        Tue, 24 Mar 1998 03:32:40 -0500 (EST)
   
   
   I think I've fixed up LessTif to do the "right thing" when an application
   is using non-default colormaps.  I've only tested the patch with mgv
   (run mgv with the -install switch to get a private colormap) but there
   shouldn't be any problems with other Motif applications.  The patch was
   made against lesstif-current from 1998.03.23.
   
   A summary of the changes follows:
   
   	include/Motif-1.2/XmI/MacrosI.h
   		Added the ColormapOfObject macro.
   
   	include/Motif-1.2/XmI/XmI.h
   		Added prototypes for the new _XmWhitePixelOfObject and
   		_XmBlackPixelOfObject functions (the implementations
   		are in Visual.c).
   
   	lib/Xm/ArrowB.c	lib/Xm/ArrowBG.c
   	lib/Xm/Label.c	lib/Xm/LabelG.c
   	lib/Xm/List.c
   		WhitePixelOfScreen and BlackPixelOfScreen changed to
   		_XmWhitePixelOfObject and _XmBlackPixelOfObject.
   
   	lib/Xm/MainW.c
   	lib/Xm/ScrollBar.c
   	lib/Xm/ScrolledW.c
   		WhitePixelOfScreen and BlackPixelOfScreen changed to
   		_XmWhitePixelOfObject and _XmBlackPixelOfObject;
   		DefaultColormapOfScreen changed to CoreColormap.
   
   	lib/Xm/Visual.c
   		WhitePixelOfScreen and BlackPixelOfScreen changed to
   		_XmWhitePixelOfObject and _XmBlackPixelOfObject;
   		DefaultColormapOfScreen changed to CoreColormap.
   		Added the _XmWhitePixel, _XmBlackPixel,
   		_XmWhitePixelOfObject, and _XmBlackPixelOfObject
   		functions.  _XmWhitePixel and _XmBlackPixel had
   		prototypes in Xm/XmP.h but no implementations,
   		_XmWhitePixelOfObject and _XmBlackPixelOfObject
   		are more convenient forms of _XmWhitePixel and
   		_XmBlackPixel.
   
   Since the drag-and-drop code draws on the root window (AFAIK), I left it
   using the default colormap.
   
   A (very) short summary of what to do when you need a Colormap or black/white
   Pixel value is at
   
   	http://www.echo-on.net/~mu/lesstif-colors.html
MainW.c 1.27
   date: 1998/03/18 21:55:14;  author: u27113;  state: Exp;  lines: +17 -2
   Vendor.c : Fix by Simon Gerraty <sjg@quick.com.au> to make Java AWT apps
   	not move to the upper left of the screen when coordinates given
   	are 0,0; it seems they pass 1,1 for that to happen.
   
   MainW.c : One of Alexander Mai's <mai@migdal.ikp.physik.tu-darmstadt.de>
   	complaints: now nedit shows the statistics line again. Somehow
   	XmMainWindow was losing some information that it did have.


Makefile.in Current rev 1.15 Last Release 1.13
Makefile.in 1.15
   date: 1997/12/26 21:44:55;  author: u27113;  state: dead;  lines: +0 -0
   Remove files.
Makefile.in 1.14
   date: 1997/12/21 14:16:52;  author: rwscott;  state: Exp;  lines: +327 -198
   This is all with respect to tearoffs.
   
   
   CascadeB.c
   
   	MenuBarEnter - Popdown the old menu only if a menushell (not torn off)
   
   	CascadePopupHandler - make sure the menu is back in its menu shell if
   		it was torn off
   
   	DelayedArm - Do not put a torn off menu back to its menushell here.
   		If the delay never times out then the menu is left in the
   		menu shell and never poped up.
   
   	CheckDisarm - If we have to popdown the menus use popdownEveryone.
   
   MenuShell.c
   
   	change_managed - If there are no managed children we pop down but, we
   	must popdown private shells different from public ones. Same goes for
   	popping up.
   
   	MenuShellPopdownOne - do not restore tear off to its toplevel shell
   	here, it just confuses popdowneveryone.
   
   	MenuShellPopdownEveryone - restore tear offs to their toplevel shells
   	here, after all the menus have been sorted out.
   
   	MenuShellPopupSharedMenuPane - Since we do not use XtPopup for menus
   	anymore we must take care of setting and un-setting of Shell_PoppedUp.
   	Not setting this was messing up <BtnDown> in push button and
   	probably other places (anywhere there was an if (Shell_PoppedUp)).
   
   PushB.c
   
   	Add some debug statements.
   
   RowColumn.c
   
   	DoBtnEventCleanupReplay - Do not call PopdownDone for a torn off menu
   
   	_XmGetActiveTopLevelMenu - Return something usefull for an active
   	tear off.
   
   TearOff.c
   
   	_XmTearOffBtnDownEventHandler - Arm the menu. Something a little
   	strange happens here, see the comments in the source.
   
   	_XmTearOffBtnUpEventHandler - DisArm the menu.
   
   	_XmTearOffInitiate - Get rid of the keyboard and pointer grabs from
   	the previous menu and turn off drag mode.
   
   	_XmAddTearOffEventHandlers - big time guess here! Add a button up
   	handler to the tear off and all of its widget children so we can
   	disarm the menu. Add a button down handler to all of the widget
   	children so we can arm the menu.
   
   	_XmRestoreTearOffToToplevelShell - Do not ungrab things here. The
   	grabs are now taken care of in the event handlers.
   	Do not reposition the shell if an event is not given.  This is due
   	the the MenuShell:change_managed call to put the menu back to the
   	toplevel shell.
   	Make sure the event handlers are in place.
   
   	_XmRestoreTearOffToMenuShell - Time for a little magic here! Ever
   	wonder how Motif keeps a copy of the tear off in the toplevel shell
   	and displays it in the menu at the same time?  Well you just take a
   	copy of the pixels in the toplevel shell before you re-parent the
   	menu and put those pixels into the shell.  Since the menu will have
   	grabs there is no way to interact with it so fake it!
   	I should be able to just set the background pixmap of the shell and
   	have X take care of the copy, stay tuned.
   	Do not unmanage the menu if it is in a private shell, they are always
   	managed.
Makefile.in 1.12
   date: 1997/08/19 19:29:12;  author: miers;  state: Exp;  lines: +5 -3
   More code reorg.
   
   1) Moved all the Xme functions to the 2.0 tree.  Something needs to be
   done about the headers, though.
   
   2) If LESSTIF_PRODUCTION is defined, then the Editres handler and the
      pixmap extensions aren't built.  I'd like to move these to another
      place (e.g., extensions/) but it's so much trouble to hack configure
      that I'm putting it off.
   
   3) Add an Editres for R5.
Makefile.in 1.11
   date: 1997/07/08 19:00:03;  author: miers;  state: Exp;  lines: +1 -0
   Moved tranlation strings back into Transltns.c.
Makefile.in 1.10
   date: 1997/07/05 11:46:51;  author: u27113;  state: Exp;  lines: +1 -0
   Added new file for (in Motif 1.2) Obsolete stuff.
Makefile.in 1.9
   date: 1997/05/27 19:46:52;  author: miers;  state: Exp;  lines: +1 -0
   A suggestion from Peter Williams: separate the strings out of misc.c
   and put them in XmStrDefs.c; this allows misc.c to be reused for both
   lib versions.
Makefile.in 1.8
   date: 1997/05/01 01:12:41;  author: miers;  state: Exp;  lines: +1 -0
   Attempting to fix RowColumn.  Peter, hopefully this won't disturb your work
   too much.
   
   RCUtils.c:  New file.  All of RowColumn's layout procedures and convenience
   functions have been moved to this file, to make RowColumn and the layout
   procedures more readable.  This should help in tracking down the layout bugs.
   
   RowColumn.c: Removed layout and convenience functions.
   
   Makefile.in: Add RCUtils to compile list.
Makefile.in 1.7
   date: 1997/04/24 18:04:38;  author: miers;  state: Exp;  lines: +2 -2
   Minor formatting changes
Makefile.in 1.6
   date: 1997/03/26 03:25:10;  author: miers;  state: Exp;  lines: +1 -1
   *: Cleaned up configuration a bit.  @exec_prefix@/bin and @exec_prefix@/lib
   become @bindir@ and @libdir@.  --with-testXm and --with-testMrm become
   booleans in test/Makefile.in.
Makefile.in 1.5
   date: 1997/03/17 22:06:04;  author: toshok;  state: Exp;  lines: +2 -1
   Need $(srcdir)/../../include in CFLAGS to fix builds in separate directories
Makefile.in 1.4
   date: 1997/03/04 07:07:04;  author: u27113;  state: Exp;  lines: +1 -0
   Form.c : trying to find a bug; added XdbDebug statements now.
   	Removed #define LESSTIF_VERBOSE which is in configure.in now.
   List.c : changed my earlier comment now that the bug with emtpy list
   	display got fixed.
   	Fixed a bug resulting in negative LIST_TopPosition (crashed ML).
   Makefile.in : added Messages.c
   Messages.c : initially generated by linking with Motif from test/Xm/misc/test14.c
   	Zmail and Netscape should link and run (somewhat) now.
   MenuShell.c : accelerators in pulldown menus didn't work any more.  Quick fix.
   
   	Mnemonics on menu bars look broken too.
   	Peter are you working still on this ?
Makefile.in 1.3
   date: 1997/02/21 08:03:13;  author: u27113;  state: Exp;  lines: +0 -9
   removed gprof junk
Makefile.in 1.2
   date: 1997/02/20 22:34:22;  author: miers;  state: Exp;  lines: +4 -2
   Moved include hierarchies so that people working on CDE/2.0/2.1 can just
   use e.g. -I../../../include/Motif-CDE -I../../../include/Motif-1.2
   to pick and choose files.
Makefile.in 1.1
   date: 1997/02/20 02:48:43;  author: miers;  state: Exp;
   It's as bad as it looks;)
   
   I HIGHLY recommend to core people that you blow away your source tree, and
   checkout with the prune (-P) option to get rid of empty directories, until
   or unless the empty directories are removed from the repository.
   =============================================================================
   
   RCS file: /cvsroot/hungry/lesstif/lib/Xm/Manager.c,v
   Working file: Manager.c
   head: 1.49
   branch:
   locks: strict
   access list:
   symbolic names:
   	release_0_83: 1.46
   	release_0_82: 1.37
   	release_0_81: 1.10
   	release_0_80public: 1.4
   	release_0_80: 1.4
   	release_0_79: 1.3
   	release_0_78: 1.3
   keyword substitution: kv
   total revisions: 49;	selected revisions: 4
   description:
Makefile.in 1.49
   date: 1998/04/16 21:46:44;  author: u27113;  state: Exp;  lines: +6 -2
   TextIn.c : don't allow user to paste into a widget that's not editable.
   	Reported by Calum Grant <Calum.Grant@cl.cam.ac.uk>.
   VirtKeys.c Primitive.c Manager.c :
   	Try to fix the fact that PageDown doesn't work in nedit.
   	The last update on VirtKeys.c inadvertently already implemented
   	a hack which makes it work.
   
   	It now looks like I found an inelegant but better solution :
   	- nedit shows that PageUp behaves like PageLeft and PageDown
   	  behaves like PageRight.
   	- a look at the virtual keys table shows that these are some
   	  of the keys having a modifier.
   	- the comments in VirtKeys.c (from Harald Albrecht ?) say that
   	  CheckForVirtualKey gets called multiple times for every key.
   	  This is not the case: this only happens the first time the user
   	  hits a key in an application. This makes me think that Xt
   	  caches some of this. Calling XtSetKeyTranslator seems to
   	  flush the cache, so it must be Xt, not Xlib.
   	- There already was an _XmVirtualKeysHandler which puts a copy
   	  of the last key pressed in the XmDisplay widget. It was never
   	  called in all of LessTif so I added it to all widgets by
   	  putting one extra line in Primitive.c and Manager.c.
   	  I don't suppose I need to do this for Gadgets as they don't
   	  get events anyway. Manager will hopefully do that for me.
   	- The info from the event (from _XmVirtualKeysHandler)
   	  contains info about which modifiers are in effect.
   	  I'm passing this instead of the usual modifiers to CheckForVirtualKey
   	  in XmTranslateKey. Not to XtTranslateKey.
   	  Not 100ure that this is right but it seems to work.
   
   Be assured that I left out all the hours of testing, digging, trying
   which led nowhere before I figured this out.
   
   In Dutch we have a term "bit-neuken" which translates quite accurately
   into bit-fucking. This is what I had to do to find this. Sigh. Poor bits,
   poor me.
Makefile.in 1.48
   date: 1998/04/16 19:19:43;  author: u27113;  state: Exp;  lines: +4 -3
   Text.c : back out change to XmTextSetString. This should fix complaints
   	about it changing the position shown when it shouldn't.
   
   Others: as indicated by Alexander Mai <mai@migdal.ikp.physik.tu-darmstadt.de>,
   	#include <LTconfig.h> should precede (almost) all other includes
   	in the sources.
Makefile.in 1.47
   date: 1998/03/14 23:21:33;  author: rwscott;  state: Exp;  lines: +14 -11
   _XmManagerInstallMnemonic() - Do not install the mnemonic if we are not
   	dealing with a Manager.  Thanks Ti. Nice little package to show the
   	problem btw.
Makefile.in 1.46
   date: 1998/03/12 20:26:25;  author: helden;  state: Exp;  lines: +3 -5
   This is a change relative to 1.43:
   
     - Removed XAllowEvents call from _XmAcceleratorHandler (as in 1.44)
   
     - Changed GrabModeSync to GrabModeAsync for both pointer and keyboard
       mode in the XtGrabKey call.
   =============================================================================
   
   RCS file: /cvsroot/hungry/lesstif/lib/Xm/MapEvent.c,v
   Working file: MapEvent.c
   head: 1.10
   branch:
   locks: strict
   access list:
   symbolic names:
   	release_0_83: 1.10
   	release_0_82: 1.6
   	release_0_81: 1.2
   	release_0_80public: 1.1
   	release_0_80: 1.1
   	release_0_79: 1.1
   	release_0_78: 1.1
   keyword substitution: kv
   total revisions: 10;	selected revisions: 1
   description:
Makefile.in 1.10
   date: 1998/01/06 00:46:22;  author: miers;  state: Exp;  lines: +3 -2
   configure.in: Change NONSTANDARD_CONVERSIONS to NONSTANDARD_CONVERTERS.
   
   mwm/Makefile.am: There is no such file as mwmlex.h
   
   lib/*/Makefile.am: A bunch of fixes (none to the 1.2 source) for building
   on Solaris.
   
   MapEvent.c, Manager.c: Missing include of <X11/Xfuncs.h> for bzero, bcopy.
   =============================================================================
   
   RCS file: /cvsroot/hungry/lesstif/lib/Xm/MenuShell.c,v
   Working file: MenuShell.c
   head: 1.67
   branch:
   locks: strict
   access list:
   symbolic names:
   	release_0_83: 1.65
   	release_0_82: 1.52
   	release_0_81: 1.30
   	release_0_80public: 1.23
   	release_0_80: 1.22
   	release_0_79: 1.9
   	release_0_78: 1.9
   keyword substitution: kv
   total revisions: 67;	selected revisions: 3
   description:
Makefile.in 1.67
   date: 1998/04/22 22:00:58;  author: u27113;  state: Exp;  lines: +19 -2
   new test to test popups in another way.
   
   MenuShell.c : make popups stay visible if they are popped up by what I call
   	a click (rapid BtnDown-BtnUp).
Makefile.in 1.66
   date: 1998/03/19 17:07:29;  author: rwscott;  state: Exp;  lines: +3 -3
   FileSB.c:_XmDirListFix() - set the preferred height of the list to its
   	current height.
   
   MenuShell.c:MenuShellPopupSharedMenuPane() - We were trying to resize an
   	object to 0x0.  This showed in the nedit open previous file menu.
Makefile.in 1.65
   date: 1998/03/09 22:32:36;  author: u27113;  state: Exp;  lines: +6 -5
   scripts/FreeBSD/Makefile.am
   	Replace the 0.81 version number in the file by something more
   	appropriate.
   
   lib/Xm/MenuShell.c
   	Finally found the cause of my ddd problem.
   	if (VAR(old) == VAR(new)) in set_values should not be == but !=.
   	Therefore, when you do
   		XmFontListFree(VAR(old));
   		VAR(new) = XmFontListFree(VAR(new));
   	you really kill a pointer in the first line that you'll copy
   	in the next line.
   
   	Replace == by != and things work again.
   =============================================================================
   
   RCS file: /cvsroot/hungry/lesstif/lib/Xm/MenuUtil.c,v
   Working file: MenuUtil.c
   head: 1.29
   branch:
   locks: strict
   access list:
   symbolic names:
   	release_0_83: 1.26
   	release_0_82: 1.10
   	release_0_81: 1.3
   	release_0_80public: 1.2
   	release_0_80: 1.2
   	release_0_79: 1.1
   	release_0_78: 1.1
   keyword substitution: kv
   total revisions: 29;	selected revisions: 4
   description:
Makefile.in 1.29
   date: 1998/05/01 15:07:02;  author: rwscott;  state: Exp;  lines: +16 -10
   MenuUtil.c:_XmMenu[Next|Prev]Item() - If there is only 1 item in the menu
   	it is obviously the next/prev item.
   	(Deleting memWidget problem)
   
   RCUtils.c:_XmRCAdjustSize() - Jon's wrapping menu problem.  This still has
   	problems with wrapping a helpWidget until I can see what is supposed
   	to happen.
   
   RowColumn.c:change_managed() - If the RC is a submenu for an option menu
   	and the memWidget for that option menu is now not managed pick a new
   	memWidget.  Somehow we need to get the CascadeButtonGadget to resize
   	to the new size of the menu still.
   
   RowColumn.c:_XmOptionCallback() - If called with w == NULL set the
   	CascadeButtonGadget labelString to ""
Makefile.in 1.28
   date: 1998/04/29 00:48:32;  author: rwscott;  state: Exp;  lines: +3 -3
   I hope I am in your will for this one Uncle George :)  Assuming it works
   of course.
   
   
   
   MenuUtil.c:_XmMenuNextItem() - expose this so I can use it in RowColumn. I do
   	not really like doing this but it is not exactly a trivial task to
   	find the next useable menu item and I already wrote this once!
   
   RowColumn.c: RC_LastSelectToplevel() holds the widget ID of the top most menu
   	in the menu heir.  It is usually set when a menu is posted to the
   	same thing as the menu that it is being posted from (confused yet, wait
   	there is more).  In the case of a pulldown that is associated to an
   	option menu we know what this will be before it is actually posted.
   	It will be the option menu that has a subMenuId of the pulldown. So,
   	if we set the RC_LastSelectToplevel of the pulldown whenever the
   	subMenuId of the option menu is set (initialize or set_values) we can
   	maintain a pointer from the pulldown back to the option menu.  We need
   	this pointer is the case where a widget in the pulldown is destroyed.
   	The destroyed widget may be the memWidget of the option menu.  If this
   	is the case, we need to pick a new memWidget.  This is where
   	_XmMenuNextItem comes in.  This back pointer is also needed with
   	respect to the TEST29 define in RowColumn that has to do with the
   	label of the current memWidget getting changed through a set_values.
   
   What I need from someone is to know what Motif picks as the new memWidget when
   the current one is deleted and what happens when all of the items in the menu
   get destroyed!  So could someone be so kind as to run rowcolumn/test34
   compiled against Motif for me and report on what happens.  It shows an
   option menu with 3 items.  Pressing the destroy button destroys the displayed
   choice.  When you do this which item does Motif now show??  When you press
   destroy 3 times what happens???
Makefile.in 1.27
   date: 1998/03/27 22:23:57;  author: miers;  state: Exp;  lines: +14 -2
   Some comments from Ramiro Estrugo about how to get Netscape to compile
   out of the box with lesstif.  This involves adding entry points for
   _XmPopup() and _XmPopdown(), as well as including DropSMgr.h in Display.h
Makefile.in 1.26
   date: 1998/03/09 14:48:47;  author: rwscott;  state: Exp;  lines: +3 -3
   _XmMenuArmItem() - special case when there is only 1 traversable item in
   	the menu.
   =============================================================================
   
   RCS file: /cvsroot/hungry/lesstif/lib/Xm/MessageB.c,v
   Working file: MessageB.c
   head: 1.20
   branch:
   locks: strict
   access list:
   symbolic names:
   	release_0_83: 1.20
   	release_0_82: 1.19
   	release_0_81: 1.15
   	release_0_80public: 1.4
   	release_0_80: 1.4
   	release_0_79: 1.1
   	release_0_78: 1.1
   keyword substitution: kv
   total revisions: 20;	selected revisions: 1
   description:
Makefile.in 1.20
   date: 1998/02/22 01:00:45;  author: rwscott;  state: Exp;  lines: +7 -3
   Mostly class structure synchronization, get rid of un-used methods, stub out
   methods that we are supposed to have, get the compress flags right....
   
   A couple of fixes to note.  BulletinBoard was not un-drawing its shadows
   when resized bigger, and the ModifyVerify patch from a few days ago was
   segfaulting on textf/test4.
   
   Now back to the menu system......
   =============================================================================
   
   RCS file: /cvsroot/hungry/lesstif/lib/Xm/Messages.c,v
   Working file: Messages.c
   head: 1.3
   branch:
   locks: strict
   access list:
   symbolic names:
   	release_0_83: 1.3
   	release_0_82: 1.3
   	release_0_81: 1.3
   	release_0_80public: 1.2
   	release_0_80: 1.2
   	release_0_79: 1.2
   	release_0_78: 1.2
   keyword substitution: kv
   total revisions: 3;	selected revisions: 1
   description:
Makefile.in 1.3
   date: 1997/07/11 04:05:14;  author: miers;  state: Exp;  lines: +193 -194
   Thru M.  Fingers tired.
   =============================================================================
   
   RCS file: /cvsroot/hungry/lesstif/lib/Xm/Mwm.c,v
   Working file: Mwm.c
   head: 1.2
   branch:
   locks: strict
   access list:
   symbolic names:
   	release_0_83: 1.2
   	release_0_82: 1.2
   	release_0_81: 1.2
   	release_0_80public: 1.1
   	release_0_80: 1.1
   	release_0_79: 1.1
   	release_0_78: 1.1
   keyword substitution: kv
   total revisions: 2;	selected revisions: 1
   description:
Makefile.in 1.2
   date: 1997/07/11 04:05:15;  author: miers;  state: Exp;  lines: +46 -35
   Thru M.  Fingers tired.
   =============================================================================
   
   RCS file: /cvsroot/hungry/lesstif/lib/Xm/Obsolete.c,v
   Working file: Obsolete.c
   head: 1.2
   branch:
   locks: strict
   access list:
   symbolic names:
   	release_0_83: 1.2
   	release_0_82: 1.2
   	release_0_81: 1.2
   keyword substitution: kv
   total revisions: 2;	selected revisions: 1
   description:
Makefile.in 1.2
   date: 1997/07/11 16:22:27;  author: miers;  state: Exp;  lines: +4 -4
   Indent.
   =============================================================================
   
   RCS file: /cvsroot/hungry/lesstif/lib/Xm/PanedW.c,v
   Working file: PanedW.c
   head: 1.35
   branch:
   locks: strict
   access list:
   symbolic names:
   	release_0_83: 1.32
   	release_0_82: 1.31
   	release_0_81: 1.28
   	release_0_80public: 1.21
   	release_0_80: 1.20
   	release_0_79: 1.9
   	release_0_78: 1.5
   keyword substitution: kv
   total revisions: 35;	selected revisions: 4
   description:
Makefile.in 1.35
   date: 1998/05/13 21:06:27;  author: u27113;  state: Exp;  lines: +14 -2
   lib/Xm/PanedW.c, lib/Xm-2.0/PanedW.c
   	Jon A. Christopher's fix.
   
   lib/Xm/TextF.c
   	Some work for using XmClipboard API (incomplete)
   
   lib/Xm/CutPaste.c
   	A couple of XdbDebugs + one core dump prevention
   
   CURRENT_NOTES
   	Comment added about using XmClipboard API in XmTextField.
Makefile.in 1.34
   date: 1998/03/26 00:38:01;  author: rwscott;  state: Exp;  lines: +7 -7
   Okay let's try this one.  Back out yesterdays patch for mainw/test12 and
   give this a try.
Makefile.in 1.33
   date: 1998/03/25 03:59:20;  author: rwscott;  state: Exp;  lines: +6 -6
   change_managed() - Although I am not an expert on PanedW's, I can rarely get
   	them to do what I want, I suspect that if the sash is not drawn then
   	the corresponding separator should not be drawn either.  This was
   	shown by mainw/test12.
Makefile.in 1.32
   date: 1998/02/22 01:00:46;  author: rwscott;  state: Exp;  lines: +15 -7
   Mostly class structure synchronization, get rid of un-used methods, stub out
   methods that we are supposed to have, get the compress flags right....
   
   A couple of fixes to note.  BulletinBoard was not un-drawing its shadows
   when resized bigger, and the ModifyVerify patch from a few days ago was
   segfaulting on textf/test4.
   
   Now back to the menu system......
   =============================================================================
   
   RCS file: /cvsroot/hungry/lesstif/lib/Xm/Primitive.c,v
   Working file: Primitive.c
   head: 1.28
   branch:
   locks: strict
   access list:
   symbolic names:
   	release_0_83: 1.27
   	release_0_82: 1.24
   	release_0_81: 1.10
   	release_0_80public: 1.6
   	release_0_80: 1.6
   	release_0_79: 1.3
   	release_0_78: 1.2
   keyword substitution: kv
   total revisions: 28;	selected revisions: 2
   description:
Makefile.in 1.28
   date: 1998/04/16 21:46:42;  author: u27113;  state: Exp;  lines: +6 -2
   TextIn.c : don't allow user to paste into a widget that's not editable.
   	Reported by Calum Grant <Calum.Grant@cl.cam.ac.uk>.
   VirtKeys.c Primitive.c Manager.c :
   	Try to fix the fact that PageDown doesn't work in nedit.
   	The last update on VirtKeys.c inadvertently already implemented
   	a hack which makes it work.
   
   	It now looks like I found an inelegant but better solution :
   	- nedit shows that PageUp behaves like PageLeft and PageDown
   	  behaves like PageRight.
   	- a look at the virtual keys table shows that these are some
   	  of the keys having a modifier.
   	- the comments in VirtKeys.c (from Harald Albrecht ?) say that
   	  CheckForVirtualKey gets called multiple times for every key.
   	  This is not the case: this only happens the first time the user
   	  hits a key in an application. This makes me think that Xt
   	  caches some of this. Calling XtSetKeyTranslator seems to
   	  flush the cache, so it must be Xt, not Xlib.
   	- There already was an _XmVirtualKeysHandler which puts a copy
   	  of the last key pressed in the XmDisplay widget. It was never
   	  called in all of LessTif so I added it to all widgets by
   	  putting one extra line in Primitive.c and Manager.c.
   	  I don't suppose I need to do this for Gadgets as they don't
   	  get events anyway. Manager will hopefully do that for me.
   	- The info from the event (from _XmVirtualKeysHandler)
   	  contains info about which modifiers are in effect.
   	  I'm passing this instead of the usual modifiers to CheckForVirtualKey
   	  in XmTranslateKey. Not to XtTranslateKey.
   	  Not 100ure that this is right but it seems to work.
   
   Be assured that I left out all the hours of testing, digging, trying
   which led nowhere before I figured this out.
   
   In Dutch we have a term "bit-neuken" which translates quite accurately
   into bit-fucking. This is what I had to do to find this. Sigh. Poor bits,
   poor me.
Makefile.in 1.27
   date: 1997/11/22 18:37:12;  author: miers;  state: Exp;  lines: +2 -4
   MacrosI.h: Add macros for Core fields when we want them not as functions.
   
   Primitive.c: I accidentally removed one of the pixmap default functions.
   
   Visual.c: Gilles Seguin tells me that "unspecified_pixmap" is a valid string
   value for the various pixmaps -- it just means XmUNSPECIFIED_PIXMAP.
   =============================================================================
   
   RCS file: /cvsroot/hungry/lesstif/lib/Xm/Protocols.c,v
   Working file: Protocols.c
   head: 1.3
   branch:
   locks: strict
   access list:
   symbolic names:
   	release_0_83: 1.3
   	release_0_82: 1.2
   	release_0_81: 1.2
   	release_0_80public: 1.1
   	release_0_80: 1.1
   	release_0_79: 1.1
   	release_0_78: 1.1
   keyword substitution: kv
   total revisions: 3;	selected revisions: 1
   description:
Makefile.in 1.3
   date: 1998/02/22 01:00:47;  author: rwscott;  state: Exp;  lines: +8 -4
   Mostly class structure synchronization, get rid of un-used methods, stub out
   methods that we are supposed to have, get the compress flags right....
   
   A couple of fixes to note.  BulletinBoard was not un-drawing its shadows
   when resized bigger, and the ModifyVerify patch from a few days ago was
   segfaulting on textf/test4.
   
   Now back to the menu system......
   =============================================================================
   
   RCS file: /cvsroot/hungry/lesstif/lib/Xm/PushB.c,v
   Working file: PushB.c
   head: 1.35
   branch:
   locks: strict
   access list:
   symbolic names:
   	release_0_83: 1.34
   	release_0_82: 1.22
   	release_0_81: 1.13
   	release_0_80public: 1.3
   	release_0_80: 1.3
   	release_0_79: 1.1
   	release_0_78: 1.1
   keyword substitution: kv
   total revisions: 35;	selected revisions: 2
   description:
Makefile.in 1.35
   date: 1998/03/16 21:52:13;  author: rwscott;  state: Exp;  lines: +12 -14
   I could not get Andrea's varargs patch to go in clean so I tried it by hand.
   
   defined the rcsid's as const
   
   cleaned out some of the unused variables.
Makefile.in 1.34
   date: 1998/02/28 18:42:00;  author: rwscott;  state: Exp;  lines: +58 -2
   The last of the menu clean up.  All menu related functions are now in RC and
   invoked through the RC_Menu class procs.  The menu item controls now install
   a version of the label menu procs tailored for them.  The menu code, in RC,
   can now highlight/unhighlight a menu control by a call to the label menu
   proc of the control.  All that is left is to get rid of several hundred lines
   of #if 0'ed out code.
   
   Now, back to keyboard traversal of menus......
   =============================================================================
   
   RCS file: /cvsroot/hungry/lesstif/lib/Xm/PushBG.c,v
   Working file: PushBG.c
   head: 1.37
   branch:
   locks: strict
   access list:
   symbolic names:
   	release_0_83: 1.33
   	release_0_82: 1.19
   	release_0_81: 1.9
   	release_0_80public: 1.4
   	release_0_80: 1.4
   	release_0_79: 1.4
   	release_0_78: 1.4
   keyword substitution: kv
   total revisions: 37;	selected revisions: 5
   description:
Makefile.in 1.37
   date: 1998/05/03 17:03:36;  author: rwscott;  state: Exp;  lines: +3 -3
   I give up on the xdir segfault problem.  Change the ArmAndActivate to
   directly call ArmTimeout instead of scheduling a delay.  This stops the
   segfault when return is pressed in the xdir remote connect text fields, with
   the downside being that we lose the visual effect of the button press.
Makefile.in 1.36
   date: 1998/03/18 17:03:19;  author: rwscott;  state: Exp;  lines: +7 -10
   ArmTimeout() - Call the Disarm action before calling the activate callbacks.
   	This keeps xdir and it's latest problem happy. It also reverses
   	the order of the activate and disarm callbacks.  This needs to be
   	checked out.
Makefile.in 1.35
   date: 1998/03/17 21:33:08;  author: rwscott;  state: Exp;  lines: +10 -2
   Arm() - I do not think the traversal should be changed here.  When the
   	ArmAndActivate action is invoked the focus does not change.  If the
   	process traversal needs to be here then ArmAndActivate cannot call
   	Arm.
Makefile.in 1.34
   date: 1998/03/16 21:52:16;  author: rwscott;  state: Exp;  lines: +12 -14
   I could not get Andrea's varargs patch to go in clean so I tried it by hand.
   
   defined the rcsid's as const
   
   cleaned out some of the unused variables.
Makefile.in 1.33
   date: 1998/03/09 04:26:47;  author: rwscott;  state: Exp;  lines: +34 -4
   Changed the set_values debug and added a comment in ArmAndActivate about
   a problem that showed up in xdir.  In summary if the time out is used
   in ArmAndActivate instead of calling ArmTimeout directly xdir will segfault
   under certain conditions.  For more details take a look at the comments in
   ArmAndActivate.
   =============================================================================
   
   RCS file: /cvsroot/hungry/lesstif/lib/Xm/RCUtils.c,v
   Working file: RCUtils.c
   head: 1.74
   branch:
   locks: strict
   access list:
   symbolic names:
   	release_0_83: 1.72
   	release_0_82: 1.68
   	release_0_81: 1.57
   	release_0_80public: 1.47
   	release_0_80: 1.47
   keyword substitution: kv
   total revisions: 74;	selected revisions: 3
   description:
Makefile.in 1.74
   date: 1998/05/01 22:51:03;  author: rwscott;  state: Exp;  lines: +64 -2
   RCUtils.c:DoLayoutHT() - treat the help button more like a regular button.
   	Still lay it out last but apply the same line wrapping as the other
   	children and, do this immediately after layout of the other children.
   	This gets the fill correct if the help button is the only widget on
   	the new row.
   
   Jon, can you let me know if this is dealing with the help button correctly.
Makefile.in 1.73
   date: 1998/05/01 15:07:03;  author: rwscott;  state: Exp;  lines: +3 -3
   MenuUtil.c:_XmMenu[Next|Prev]Item() - If there is only 1 item in the menu
   	it is obviously the next/prev item.
   	(Deleting memWidget problem)
   
   RCUtils.c:_XmRCAdjustSize() - Jon's wrapping menu problem.  This still has
   	problems with wrapping a helpWidget until I can see what is supposed
   	to happen.
   
   RowColumn.c:change_managed() - If the RC is a submenu for an option menu
   	and the memWidget for that option menu is now not managed pick a new
   	memWidget.  Somehow we need to get the CascadeButtonGadget to resize
   	to the new size of the menu still.
   
   RowColumn.c:_XmOptionCallback() - If called with w == NULL set the
   	CascadeButtonGadget labelString to ""
Makefile.in 1.72
   date: 1998/01/22 23:14:10;  author: u27113;  state: Exp;  lines: +7 -2
   Some updates to documents; HTML errors, a forgotten update to
   our version history, adding wxWindows.
   
   RCUtils : add XdbDebug trying to find Jon Christopher's problem.
   	Nothing yet.
   =============================================================================
   
   RCS file: /cvsroot/hungry/lesstif/lib/Xm/ReadImage.c,v
   Working file: ReadImage.c
   head: 1.1
   branch:
   locks: strict
   access list:
   symbolic names:
   	release_0_83: 1.1
   	release_0_82: 1.1
   	release_0_81: 1.1
   	release_0_80public: 1.1
   	release_0_80: 1.1
   	release_0_79: 1.1
   	release_0_78: 1.1
   keyword substitution: kv
   total revisions: 1;	selected revisions: 1
   description:
Makefile.in 1.1
   date: 1997/02/20 02:48:55;  author: miers;  state: Exp;
   It's as bad as it looks;)
   
   I HIGHLY recommend to core people that you blow away your source tree, and
   checkout with the prune (-P) option to get rid of empty directories, until
   or unless the empty directories are removed from the repository.
   =============================================================================
   
   RCS file: /cvsroot/hungry/lesstif/lib/Xm/Region.c,v
   Working file: Region.c
   head: 1.3
   branch:
   locks: strict
   access list:
   symbolic names:
   	release_0_83: 1.3
   	release_0_82: 1.2
   	release_0_81: 1.2
   	release_0_80public: 1.1
   	release_0_80: 1.1
   	release_0_79: 1.1
   	release_0_78: 1.1
   keyword substitution: kv
   total revisions: 3;	selected revisions: 1
   description:
Makefile.in 1.3
   date: 1998/02/11 21:15:12;  author: u27113;  state: Exp;  lines: +8 -4
   lib/Xm/Makefile.am, lib/Mrm/Makefile.am :
   	fix version numbers on Linux
   	Libtool has a rather strange way of calculating them on Linux,
   	it's more normal for FreeBSD so I never noticed.
   
   lib/Mrm/Region.c :
   	remove memory leak as indicated by Alan Curry <pacman@cqc.com>
   
   doc/www.lesstif.org/working.html :
   	Change status of Xquote. Its web site says it works.
   
   scripts/mmf :
   	More intelligent version.
   
   CURRENT_NOTES : change comment for mmf slightly
   
   include/Motif-1.2/Xm/Makefile.am, include/Motif-2.0/Xm/Makefile.am :
   	Make sure Xm.h (which is built by configure and may therefore
   	be in a build directory tree which is potentially different
   	from the source directory tree) gets installed right.
   =============================================================================
   
   RCS file: /cvsroot/hungry/lesstif/lib/Xm/RepType.c,v
   Working file: RepType.c
   head: 1.3
   branch:
   locks: strict
   access list:
   symbolic names:
   	release_0_83: 1.3
   	release_0_82: 1.3
   	release_0_81: 1.2
   	release_0_80public: 1.1
   	release_0_80: 1.1
   	release_0_79: 1.1
   	release_0_78: 1.1
   keyword substitution: kv
   total revisions: 3;	selected revisions: 1
   description:
Makefile.in 1.3
   date: 1997/11/17 19:53:38;  author: u27113;  state: Exp;  lines: +7 -2
   OS/2 fixes that I forgot about for a while.
   =============================================================================
   
   RCS file: /cvsroot/hungry/lesstif/lib/Xm/ResConvert.c,v
   Working file: ResConvert.c
   head: 1.25
   branch:
   locks: strict
   access list:
   symbolic names:
   	release_0_83: 1.20
   	release_0_82: 1.15
   	release_0_81: 1.10
   	release_0_80public: 1.3
   	release_0_80: 1.3
   	release_0_79: 1.2
   	release_0_78: 1.2
   keyword substitution: kv
   total revisions: 25;	selected revisions: 6
   description:
Makefile.in 1.25
   date: 1998/04/23 22:27:32;  author: u27113;  state: Exp;  lines: +6 -2
   Moved obsolete OS/2 configuration file to scripts/OS2.
   
   ExtObjectP.h : as <gatgul@voicenet.com> suggested, make XmNUM_BYTES
           such that XmExtCache has a size which is a multiple of the
           size of a pointer. Used a simpler definition though.
           This should (theoretically) be better for performance on Alpha.
   
   configure.in, lib/Xm/ResConvert.c : solve strtoul problem for SunOS 4.1
           How come people are still using this. Hasn't SunOS 4.1 been
           obsoleted about 6 years ago? Oh well.
   
   RenderTable.c Xm.h.in : add stuff for 2.0's XmRender* API's.
           All empty of course :-(
Makefile.in 1.24
   date: 1998/04/16 19:19:41;  author: u27113;  state: Exp;  lines: +8 -3
   Text.c : back out change to XmTextSetString. This should fix complaints
   	about it changing the position shown when it shouldn't.
   
   Others: as indicated by Alexander Mai <mai@migdal.ikp.physik.tu-darmstadt.de>,
   	#include <LTconfig.h> should precede (almost) all other includes
   	in the sources.
Makefile.in 1.23
   date: 1998/04/11 14:52:30;  author: u27113;  state: Exp;  lines: +78 -12
   lib/Xm/ResConvert.c
   	A better implementation that was submitted by Alex Lukyanov a long
   	time ago, but overlooked.
   lib/Xm/Text.c
   	Fix nedit's status line: XmTextSetString and XmTextReplace should
   	show the last position that they modified.
Makefile.in 1.22
   date: 1998/04/07 20:38:16;  author: u27113;  state: Exp;  lines: +35 -2
   Vendor.c : fix for 64-bit systems from Alexander V. Lukyanov <lav@yars.free.net>
   ResConvert.c : Add a converter for Xinvest.
   XmStrDefs.h : add XmNlist for Mozilla (only if Motif 2.0).
   lib/Xm/TextF.c : fix by Kevin B. Hendricks <kbhend@dogwood.tyler.wm.edu>
Makefile.in 1.21
   date: 1998/03/17 19:59:31;  author: u27113;  state: Exp;  lines: +65 -4
   Fixes by Andreas Zeller
Makefile.in 1.20
   date: 1998/03/14 15:00:06;  author: u27113;  state: Exp;  lines: +54 -3
   Put the 2.0 stuff into lib/Xm/ResConvert.c, so we can get rid of
   lib/Xm-2.0/ResConvert.c.
   Change the build so the right file is picked up.
   Remove lib/Xm-2.0/ResConvert.c .
   =============================================================================
   
   RCS file: /cvsroot/hungry/lesstif/lib/Xm/ResInd.c,v
   Working file: ResInd.c
   head: 1.4
   branch:
   locks: strict
   access list:
   symbolic names:
   	release_0_83: 1.4
   	release_0_82: 1.4
   	release_0_81: 1.4
   	release_0_80public: 1.2
   	release_0_80: 1.2
   	release_0_79: 1.1
   	release_0_78: 1.1
   keyword substitution: kv
   total revisions: 4;	selected revisions: 1
   description:
Makefile.in 1.4
   date: 1997/08/19 19:29:14;  author: miers;  state: Exp;  lines: +2 -38
   More code reorg.
   
   1) Moved all the Xme functions to the 2.0 tree.  Something needs to be
   done about the headers, though.
   
   2) If LESSTIF_PRODUCTION is defined, then the Editres handler and the
      pixmap extensions aren't built.  I'd like to move these to another
      place (e.g., extensions/) but it's so much trouble to hack configure
      that I'm putting it off.
   
   3) Add an Editres for R5.
   =============================================================================
   
   RCS file: /cvsroot/hungry/lesstif/lib/Xm/RowColumn.c,v
   Working file: RowColumn.c
   head: 1.147
   branch:
   locks: strict
   access list:
   symbolic names:
   	release_0_83: 1.138
   	release_0_82: 1.103
   	release_0_81: 1.72
   	release_0_80public: 1.63
   	release_0_80: 1.63
   	release_0_79: 1.25
   	release_0_78: 1.22
   keyword substitution: kv
   total revisions: 147;	selected revisions: 10
   description:
Makefile.in 1.147
   date: 1998/05/07 23:06:14;  author: rwscott;  state: Exp;  lines: +8 -2
   _XmPostPopupMenu() - Mozilla's Bookmarks, Back and, Forward buttons have
   	menus attached, however when we get to posting them, somehow, they
   	are not managed.  This assumes that the menu to be posted should also
   	be managed and therefore calls XtManageChild before posting.
   
   This gets Mozilla's Bookmarks, Back and, Forward menus posting.
   
   Next, I don't suppose the pixmaps in the buttons is supposed to overlap the
   text.  Problem is I have not been able to find these buttons yet!!
Makefile.in 1.146
   date: 1998/05/05 21:32:34;  author: rwscott;  state: Exp;  lines: +45 -12
   Just some comments wrt the menu system (try to figure out how it works again)
Makefile.in 1.145
   date: 1998/05/05 03:19:28;  author: rwscott;  state: Exp;  lines: +6 -2
   realize() - I may have just found out why RC has a realize method.  It is
   	there for Mozilla's Go menu!   When Mozilla updates its Go menu
   	it first UnRealizes it, who knows why, and then Realizes it.  The
   	UnRealize also UnManages it.  The problem is that a menu bar utilizes
   	a shared menu shell.  Because of this any menus added to a menushell
   	get managed in the MenuShell insert_child method.  Since Mozilla is
   	un-realizing the menu, and therefore un-managing it, the menu never
   	gets managed again and, therefore the menu bar is royally messed up!
   	So, if a RowColumn is being realized and it is in a private menu shell
   	lets manage the thing and see what happens.  It seems to work for
   	Mozilla and does not bother anything else so......
Makefile.in 1.144
   date: 1998/05/01 15:07:04;  author: rwscott;  state: Exp;  lines: +26 -15
   MenuUtil.c:_XmMenu[Next|Prev]Item() - If there is only 1 item in the menu
   	it is obviously the next/prev item.
   	(Deleting memWidget problem)
   
   RCUtils.c:_XmRCAdjustSize() - Jon's wrapping menu problem.  This still has
   	problems with wrapping a helpWidget until I can see what is supposed
   	to happen.
   
   RowColumn.c:change_managed() - If the RC is a submenu for an option menu
   	and the memWidget for that option menu is now not managed pick a new
   	memWidget.  Somehow we need to get the CascadeButtonGadget to resize
   	to the new size of the menu still.
   
   RowColumn.c:_XmOptionCallback() - If called with w == NULL set the
   	CascadeButtonGadget labelString to ""
Makefile.in 1.143
   date: 1998/04/29 00:48:33;  author: rwscott;  state: Exp;  lines: +56 -8
   I hope I am in your will for this one Uncle George :)  Assuming it works
   of course.
   
   
   
   MenuUtil.c:_XmMenuNextItem() - expose this so I can use it in RowColumn. I do
   	not really like doing this but it is not exactly a trivial task to
   	find the next useable menu item and I already wrote this once!
   
   RowColumn.c: RC_LastSelectToplevel() holds the widget ID of the top most menu
   	in the menu heir.  It is usually set when a menu is posted to the
   	same thing as the menu that it is being posted from (confused yet, wait
   	there is more).  In the case of a pulldown that is associated to an
   	option menu we know what this will be before it is actually posted.
   	It will be the option menu that has a subMenuId of the pulldown. So,
   	if we set the RC_LastSelectToplevel of the pulldown whenever the
   	subMenuId of the option menu is set (initialize or set_values) we can
   	maintain a pointer from the pulldown back to the option menu.  We need
   	this pointer is the case where a widget in the pulldown is destroyed.
   	The destroyed widget may be the memWidget of the option menu.  If this
   	is the case, we need to pick a new memWidget.  This is where
   	_XmMenuNextItem comes in.  This back pointer is also needed with
   	respect to the TEST29 define in RowColumn that has to do with the
   	label of the current memWidget getting changed through a set_values.
   
   What I need from someone is to know what Motif picks as the new memWidget when
   the current one is deleted and what happens when all of the items in the menu
   get destroyed!  So could someone be so kind as to run rowcolumn/test34
   compiled against Motif for me and report on what happens.  It shows an
   option menu with 3 items.  Pressing the destroy button destroys the displayed
   choice.  When you do this which item does Motif now show??  When you press
   destroy 3 times what happens???
Makefile.in 1.142
   date: 1998/04/27 20:44:40;  author: rwscott;  state: Exp;  lines: +3 -3
   Damn, that did not last long!  Only check to see if the MemWidget is
   managed if the MemWidget exists in the first place!
Makefile.in 1.141
   date: 1998/04/27 20:30:11;  author: rwscott;  state: Exp;  lines: +7 -2
   change_managed() - If the RC_MemWidget is not managed, pick a new one by
   	calling _XmFixOptionMenu.
   
   This should take care of Uncle George's problem of un-managing and
   destroying the active item in an option menu.
Makefile.in 1.140
   date: 1998/04/02 22:28:27;  author: rwscott;  state: Exp;  lines: +9 -2
   _XmPostPopupMenu() - Make the change that Andreas suggested a while back to
   	make popups work the first time they are popped up.
   	   This makes the popup's in ddd and Mozilla work but, the real
   	   problem is the GrabButton is not getting called for some reason.
   	   The grab is placed on the RC in AddPopupHandlers but it never
   	   gets triggered?????
Makefile.in 1.139
   date: 1998/03/28 02:24:53;  author: rwscott;  state: Exp;  lines: +7 -4
   _XmFixOptionMenu() - While trying to assign the MemWidget to the first item
   	in the menu, we use the second item if the first is a tear off
   	control.  This causes a problem if the tear off control is the only
   	item in the menu.  Therefore check for at least two items before
   	setting the memWidget.  This shows up now because the tear-off
   	control is being created in initialize instead of when the menu
   	is posted.
   
   
   This is for the one that Jon just reported.
Makefile.in 1.138
   date: 1998/03/13 22:25:33;  author: rwscott;  state: Exp;  lines: +10 -4
   RC_Procs - MENU_SHELL_POPDOWN - Get rid of the warning that Alexander has
   	been commenting on for a while.  I do'nt think it is entirly
   	correct but, until I find out whether or not RC_PopupPosted stays
   	set when a menu gets torn off it will do.
   
   Also added a comment around where a shell gets popped up.  With respect to
   tear-offs we have to add a grab back in when the menu gets posted back to
   the menu-shell from the tear off. I am not real comfortable leaving this in,
   but I do not want to lose it, so it is there commented out.
   =============================================================================
   
   RCS file: /cvsroot/hungry/lesstif/lib/Xm/Sash.c,v
   Working file: Sash.c
   head: 1.10
   branch:
   locks: strict
   access list:
   symbolic names:
   	release_0_83: 1.10
   	release_0_82: 1.9
   	release_0_81: 1.4
   	release_0_80public: 1.1
   	release_0_80: 1.1
   	release_0_79: 1.1
   	release_0_78: 1.1
   keyword substitution: kv
   total revisions: 10;	selected revisions: 1
   description:
Makefile.in 1.10
   date: 1998/02/22 01:00:49;  author: rwscott;  state: Exp;  lines: +10 -27
   Mostly class structure synchronization, get rid of un-used methods, stub out
   methods that we are supposed to have, get the compress flags right....
   
   A couple of fixes to note.  BulletinBoard was not un-drawing its shadows
   when resized bigger, and the ModifyVerify patch from a few days ago was
   segfaulting on textf/test4.
   
   Now back to the menu system......
   =============================================================================
   
   RCS file: /cvsroot/hungry/lesstif/lib/Xm/Scale.c,v
   Working file: Scale.c
   head: 1.25
   branch:
   locks: strict
   access list:
   symbolic names:
   	release_0_83: 1.23
   	release_0_82: 1.22
   	release_0_81: 1.14
   	release_0_80public: 1.4
   	release_0_80: 1.3
   	release_0_79: 1.1
   	release_0_78: 1.1
   keyword substitution: kv
   total revisions: 25;	selected revisions: 3
   description:
Makefile.in 1.25
   date: 1998/04/22 19:47:18;  author: u27113;  state: Exp;  lines: +14 -2
   Scale.c, MacrosI.h, ScaleP.h, Makefile.am :
   	First cut at implementing Motif 2.0's XmNeditable for XmScale.
   	Ermanno Scaglione <e.scaglione@areacom.it> asked for this.
   
   	XmTHERMOMETHER is a special case of XmScale that should be
   	added to XmScale too. I didn't do that now :-)
Makefile.in 1.24
   date: 1998/03/30 21:39:20;  author: rwscott;  state: Exp;  lines: +7 -4
   This fixes Jon's problem with the scales in the Form.  The scales were
   left justifying everything after a realize instead of right.
Makefile.in 1.23
   date: 1998/02/22 01:00:50;  author: rwscott;  state: Exp;  lines: +4 -4
   Mostly class structure synchronization, get rid of un-used methods, stub out
   methods that we are supposed to have, get the compress flags right....
   
   A couple of fixes to note.  BulletinBoard was not un-drawing its shadows
   when resized bigger, and the ModifyVerify patch from a few days ago was
   segfaulting on textf/test4.
   
   Now back to the menu system......
   =============================================================================
   
   RCS file: /cvsroot/hungry/lesstif/lib/Xm/Screen.c,v
   Working file: Screen.c
   head: 1.10
   branch:
   locks: strict
   access list:
   symbolic names:
   	release_0_83: 1.10
   	release_0_82: 1.9
   	release_0_81: 1.7
   	release_0_80public: 1.4
   	release_0_80: 1.4
   	release_0_79: 1.3
   	release_0_78: 1.3
   keyword substitution: kv
   total revisions: 10;	selected revisions: 1
   description:
Makefile.in 1.10
   date: 1998/02/22 01:00:51;  author: rwscott;  state: Exp;  lines: +7 -14
   Mostly class structure synchronization, get rid of un-used methods, stub out
   methods that we are supposed to have, get the compress flags right....
   
   A couple of fixes to note.  BulletinBoard was not un-drawing its shadows
   when resized bigger, and the ModifyVerify patch from a few days ago was
   segfaulting on textf/test4.
   
   Now back to the menu system......
   =============================================================================
   
   RCS file: /cvsroot/hungry/lesstif/lib/Xm/ScrollBar.c,v
   Working file: ScrollBar.c
   head: 1.23
   branch:
   locks: strict
   access list:
   symbolic names:
   	release_0_83: 1.21
   	release_0_82: 1.16
   	release_0_81: 1.12
   	release_0_80public: 1.7
   	release_0_80: 1.7
   	release_0_79: 1.5
   	release_0_78: 1.5
   keyword substitution: kv
   total revisions: 23;	selected revisions: 3
   description:
Makefile.in 1.23
   date: 1998/04/10 19:10:44;  author: jonf;  state: Exp;  lines: +5 -3
   Changed default value of Prim_HighlightThickness to be 0,
   to match M*tif spec.
Makefile.in 1.22
   date: 1998/03/24 23:01:06;  author: rwscott;  state: Exp;  lines: +7 -8
   From:        Eric Howe <mu@wintermute.echo-on.net>
   To:          lesstif@Hungry.COM
   Subject:     Colormap patch
   Date:        Tue, 24 Mar 1998 03:32:40 -0500 (EST)
   
   
   I think I've fixed up LessTif to do the "right thing" when an application
   is using non-default colormaps.  I've only tested the patch with mgv
   (run mgv with the -install switch to get a private colormap) but there
   shouldn't be any problems with other Motif applications.  The patch was
   made against lesstif-current from 1998.03.23.
   
   A summary of the changes follows:
   
   	include/Motif-1.2/XmI/MacrosI.h
   		Added the ColormapOfObject macro.
   
   	include/Motif-1.2/XmI/XmI.h
   		Added prototypes for the new _XmWhitePixelOfObject and
   		_XmBlackPixelOfObject functions (the implementations
   		are in Visual.c).
   
   	lib/Xm/ArrowB.c	lib/Xm/ArrowBG.c
   	lib/Xm/Label.c	lib/Xm/LabelG.c
   	lib/Xm/List.c
   		WhitePixelOfScreen and BlackPixelOfScreen changed to
   		_XmWhitePixelOfObject and _XmBlackPixelOfObject.
   
   	lib/Xm/MainW.c
   	lib/Xm/ScrollBar.c
   	lib/Xm/ScrolledW.c
   		WhitePixelOfScreen and BlackPixelOfScreen changed to
   		_XmWhitePixelOfObject and _XmBlackPixelOfObject;
   		DefaultColormapOfScreen changed to CoreColormap.
   
   	lib/Xm/Visual.c
   		WhitePixelOfScreen and BlackPixelOfScreen changed to
   		_XmWhitePixelOfObject and _XmBlackPixelOfObject;
   		DefaultColormapOfScreen changed to CoreColormap.
   		Added the _XmWhitePixel, _XmBlackPixel,
   		_XmWhitePixelOfObject, and _XmBlackPixelOfObject
   		functions.  _XmWhitePixel and _XmBlackPixel had
   		prototypes in Xm/XmP.h but no implementations,
   		_XmWhitePixelOfObject and _XmBlackPixelOfObject
   		are more convenient forms of _XmWhitePixel and
   		_XmBlackPixel.
   
   Since the drag-and-drop code draws on the root window (AFAIK), I left it
   using the default colormap.
   
   A (very) short summary of what to do when you need a Colormap or black/white
   Pixel value is at
   
   	http://www.echo-on.net/~mu/lesstif-colors.html
Makefile.in 1.21
   date: 1998/03/10 20:20:23;  author: helden;  state: Exp;  lines: +2 -7
   Fix by Dan <oramd@cs.man.ac.uk>. XmScrollBarSetValues now positions the slider
   correctly.
   =============================================================================
   
   RCS file: /cvsroot/hungry/lesstif/lib/Xm/ScrolledW.c,v
   Working file: ScrolledW.c
   head: 1.76
   branch:
   locks: strict
   access list:
   symbolic names:
   	release_0_83: 1.68
   	release_0_82: 1.67
   	release_0_81: 1.61
   	release_0_80public: 1.21
   	release_0_80: 1.18
   	release_0_79: 1.11
   	release_0_78: 1.9
   keyword substitution: kv
   total revisions: 76;	selected revisions: 9
   description:
Makefile.in 1.76
   date: 1998/04/10 19:12:41;  author: jonf;  state: Exp;  lines: +27 -16
   Removed some mgv hacks that were no longer needed which
   were causing geometry problems for scrolled lists.
   
   Also had to re-enable some code that had recently be ifdef'd out
   because doing so made the mgv Page List too wide.
Makefile.in 1.75
   date: 1998/04/10 09:23:53;  author: u27113;  state: Exp;  lines: +2 -8
   doc/lessdox/widgets/XmFileSelectionBox.3
   	A small amount of real documentation.
   lib/Xm/ScrolledW.c
   	Remove yesterday's hack.
   lib/Xm/GeoUtils.c
   	To put it here :-)
   
   	In _XmMakeGeometryRequest, if we will be requesting a 0 width
   	or height, change that to 1 to prevent Xt from doing strange
   	things (like X errors).
   
   	This fixes Alexander Mai's xacc problem (message dated 6 April).
   	It also fixes the scrolling lists in selection boxes that
   	behaved rather strange lately.
   	Doesn't fix Xinvest yet.
Makefile.in 1.74
   date: 1998/04/09 12:23:45;  author: u27113;  state: Exp;  lines: +8 -2
   lib/Xm/GadgetUtil.c : Rename _XmInputInGadget for 2.0 for Mozilla.
   lib/Xm/ScrolledW.c : Don't ask for new geometry if not realized.
   	Maybe this fixes Alexander Mai's XACC problem.
   lib/Xm/TextF.c : Clean up some junk.
   lib/Xm/misc.c : Mozilla build script looks for an identifying string
   	in the shared library (in xmversion.sh).
   doc/lessdox/Makefile.am : add subdirectory widgets
Makefile.in 1.73
   date: 1998/03/29 21:57:07;  author: rwscott;  state: Exp;  lines: +4 -2
   GeoUtils.c:_XmHandleGeometryManager() - If the reisize policy permits the
   	request we check to make sure that the instigator got its requested
   	size. If it did we request the new size from the parent. However
   	we had a ! in the test so that we would request a size change from
   	the parent only if the instigator did not get the size it wanted. So
   	we take out the !.  If the instigator did not get the size it wanted
   	return Almost.  This gives a second chance in case only 1 of the
   	dimensions was not as requested.  For example if a menu bar changes
   	its height, the new height may be honoured but a MessageBox will set
   	the width to the width of the box.  Returning Almost in this case
   	gives it a second chance.
   
   ScrolledW.c:_XmScrolledWGeomRequest() - Fixing the GeoUtils allows this ugly
   	MFM_BUGS hack to be removed, I think.
Makefile.in 1.72
   date: 1998/03/28 16:10:14;  author: rwscott;  state: Exp;  lines: +50 -13
   Attempt to get rid of some of the scrollbar warnings.  In one case we were
   setting the max and min, if the old max used to be larger then there was
   a chance that the old value was also larger, which gave rise to the
   "value larger than max - slider size" warning.
Makefile.in 1.71
   date: 1998/03/25 23:44:31;  author: rwscott;  state: Exp;  lines: +28 -9
   undef MFM_BUGS again.  This along with a change in the parameters to the
   layout routine in geometry_manager.  Also a bit of a width tweak to get the
   width of the list in mgv to what I think it should be.  The tweak turns out
   to be half the value of the fudge that I had to put into FileSB when
   calculating the width of the list. What's missing?????
   
   This gets the long standing layout bug in mfm fixed as well as the page list
   in mgv.
Makefile.in 1.70
   date: 1998/03/25 21:09:49;  author: u27113;  state: Exp;  lines: +19 -2
   FontList.c : remove some memory leaks;
   	more importantly: don't return our argument when NULL passed
   	in XmFontListAppendEntry; this doesn't work with the memory
   	allocation rules.
   
   TextF.c (both) : watch out more carefully about when you XmFontListFree.
   	Yesterday's implementation did one where it shouldn't.
   	Also do whatever's necessary when changing the XmNfontList
   	through SetValues. Now the newly adapted test/textf/test2.c works.
   
   test/textf/test2.c : call setvalues on XmNfontList to show what didn't work.
   
   ScrolledW.c: add some XdbDebugs, also strip a couple of bits off what
   	GeometryManager returns.
Makefile.in 1.69
   date: 1998/03/24 23:01:10;  author: rwscott;  state: Exp;  lines: +5 -6
   From:        Eric Howe <mu@wintermute.echo-on.net>
   To:          lesstif@Hungry.COM
   Subject:     Colormap patch
   Date:        Tue, 24 Mar 1998 03:32:40 -0500 (EST)
   
   
   I think I've fixed up LessTif to do the "right thing" when an application
   is using non-default colormaps.  I've only tested the patch with mgv
   (run mgv with the -install switch to get a private colormap) but there
   shouldn't be any problems with other Motif applications.  The patch was
   made against lesstif-current from 1998.03.23.
   
   A summary of the changes follows:
   
   	include/Motif-1.2/XmI/MacrosI.h
   		Added the ColormapOfObject macro.
   
   	include/Motif-1.2/XmI/XmI.h
   		Added prototypes for the new _XmWhitePixelOfObject and
   		_XmBlackPixelOfObject functions (the implementations
   		are in Visual.c).
   
   	lib/Xm/ArrowB.c	lib/Xm/ArrowBG.c
   	lib/Xm/Label.c	lib/Xm/LabelG.c
   	lib/Xm/List.c
   		WhitePixelOfScreen and BlackPixelOfScreen changed to
   		_XmWhitePixelOfObject and _XmBlackPixelOfObject.
   
   	lib/Xm/MainW.c
   	lib/Xm/ScrollBar.c
   	lib/Xm/ScrolledW.c
   		WhitePixelOfScreen and BlackPixelOfScreen changed to
   		_XmWhitePixelOfObject and _XmBlackPixelOfObject;
   		DefaultColormapOfScreen changed to CoreColormap.
   
   	lib/Xm/Visual.c
   		WhitePixelOfScreen and BlackPixelOfScreen changed to
   		_XmWhitePixelOfObject and _XmBlackPixelOfObject;
   		DefaultColormapOfScreen changed to CoreColormap.
   		Added the _XmWhitePixel, _XmBlackPixel,
   		_XmWhitePixelOfObject, and _XmBlackPixelOfObject
   		functions.  _XmWhitePixel and _XmBlackPixel had
   		prototypes in Xm/XmP.h but no implementations,
   		_XmWhitePixelOfObject and _XmBlackPixelOfObject
   		are more convenient forms of _XmWhitePixel and
   		_XmBlackPixel.
   
   Since the drag-and-drop code draws on the root window (AFAIK), I left it
   using the default colormap.
   
   A (very) short summary of what to do when you need a Colormap or black/white
   Pixel value is at
   
   	http://www.echo-on.net/~mu/lesstif-colors.html
Makefile.in 1.68
   date: 1998/02/22 01:00:53;  author: rwscott;  state: Exp;  lines: +14 -14
   Mostly class structure synchronization, get rid of un-used methods, stub out
   methods that we are supposed to have, get the compress flags right....
   
   A couple of fixes to note.  BulletinBoard was not un-drawing its shadows
   when resized bigger, and the ModifyVerify patch from a few days ago was
   segfaulting on textf/test4.
   
   Now back to the menu system......
   =============================================================================
   
   RCS file: /cvsroot/hungry/lesstif/lib/Xm/Selection.c,v
   Working file: Selection.c
   head: 1.1
   branch:
   locks: strict
   access list:
   symbolic names:
   	release_0_83: 1.1
   	release_0_82: 1.1
   	release_0_81: 1.1
   	release_0_80public: 1.1
   	release_0_80: 1.1
   	release_0_79: 1.1
   	release_0_78: 1.1
   keyword substitution: kv
   total revisions: 1;	selected revisions: 1
   description:
Makefile.in 1.1
   date: 1997/02/20 02:49:08;  author: miers;  state: Exp;
   It's as bad as it looks;)
   
   I HIGHLY recommend to core people that you blow away your source tree, and
   checkout with the prune (-P) option to get rid of empty directories, until
   or unless the empty directories are removed from the repository.
   =============================================================================
   
   RCS file: /cvsroot/hungry/lesstif/lib/Xm/SelectionBox.c,v
   Working file: SelectionBox.c
   head: 1.28
   branch:
   locks: strict
   access list:
   symbolic names:
   	release_0_83: 1.24
   	release_0_82: 1.20
   	release_0_81: 1.18
   	release_0_80public: 1.8
   	release_0_80: 1.8
   	release_0_79: 1.3
   	release_0_78: 1.2
   keyword substitution: kv
   total revisions: 28;	selected revisions: 5
   description:
Makefile.in 1.28
   date: 1998/04/15 00:20:03;  author: rwscott;  state: Exp;  lines: +11 -2
   SelectionBox.c:initialize() - We were calling set_values here to set the
   	default button.  This meant calling set_values on a widget that is
   	not yet fully initialized.  This was probably not a big deal for
   	SelectionBox but, sub-classes have not even had there initialize
   	method called yet!  In the case of FSB the set_values was being
   	called before the initialize, which meant most of the FSB had not
   	been built yet!  Use the BB_ macro to set the default button since
   	it is a sub-class of BB anyway.
   
   This fixes the recent core dumps in the filesb tests. This also seems to fix
   the "no shadow on the menubars" in some of the FSB tests also.
   
   Xmos.c:_XmOSBuildFileList() - wrap another fprintf in an XdbDebug
Makefile.in 1.27
   date: 1998/04/14 22:54:49;  author: rwscott;  state: Exp;  lines: +19 -3
   FileSB.c LabelG.c SelectionBox.c XmString.c Xmos.c
   
   Patch from Peter Wainwright <prw@wainpr.demon.co.uk> dealing mainly with
   memory leaks during string handling.
   
   List.c:_XmListSetGeometry() - If the parent is not managed we were not asking
   	for a geo change.  The result was that a change of visibleItems while
   	the parent was not managed was not changing the visible items.  If the
   	parent is not managed Xt should take care of the size request by
   	just granting it.  This was messing up one of my apps.  If someone
   	can point out why it was there I will re-evaluate this.
   
   BulletinBoard.c:handle_change_managed() - If we are not realized, layout with
   	the current width/height, this will honor the specified resources if
   	any.
Makefile.in 1.26
   date: 1998/03/24 21:51:21;  author: rwscott;  state: Exp;  lines: +3 -3
   The problem is a failure to match a valid selection with a list item in an
   XmFileSelectionBox with XmNmustMatch set True.  The reason is when the string
   from the selection XmTextField is converted to an XmString the tag is not the
   same one used to create the list items.  Here is the fix for version 0.83.
   
   Carl Howes <cbh@ircpc-8617.msd.ray.com>
Makefile.in 1.25
   date: 1998/03/18 19:43:13;  author: u27113;  state: Exp;  lines: +2 -3
   lib/Xm/FontList.c : correction by Kevin B. Hendricks for a fix for the
   	JDK port to Linux. <kbhend@dogwood.tyler.wm.edu>
   
   lib/Xm/SelectionBox.c : fix by Andreas Zeller <zeller@acm.org>
Makefile.in 1.24
   date: 1998/03/12 23:45:21;  author: rwscott;  state: Exp;  lines: +8 -2
   insert_child() - Before ParentProcessActivate we needed to add an activate
   	callback to all the TextField children to get the behaviour correct.
   	Now that TextField calls ParentProcessActivate in response to the
   	activate event this extra callback was causing the dialog okCallback
   	to get called twice.
   
   This may have been the problem with the nedit save as function, since
   FileSelectionBox is a subclass of SelectionBox (I did not check).
   =============================================================================
   
   RCS file: /cvsroot/hungry/lesstif/lib/Xm/SeparatoG.c,v
   Working file: SeparatoG.c
   head: 1.7
   branch:
   locks: strict
   access list:
   symbolic names:
   	release_0_83: 1.7
   	release_0_82: 1.5
   	release_0_81: 1.5
   	release_0_80public: 1.3
   	release_0_80: 1.3
   	release_0_79: 1.2
   	release_0_78: 1.2
   keyword substitution: kv
   total revisions: 7;	selected revisions: 1
   description:
Makefile.in 1.7
   date: 1998/02/22 01:00:55;  author: rwscott;  state: Exp;  lines: +21 -22
   Mostly class structure synchronization, get rid of un-used methods, stub out
   methods that we are supposed to have, get the compress flags right....
   
   A couple of fixes to note.  BulletinBoard was not un-drawing its shadows
   when resized bigger, and the ModifyVerify patch from a few days ago was
   segfaulting on textf/test4.
   
   Now back to the menu system......
   =============================================================================
   
   RCS file: /cvsroot/hungry/lesstif/lib/Xm/Separator.c,v
   Working file: Separator.c
   head: 1.6
   branch:
   locks: strict
   access list:
   symbolic names:
   	release_0_83: 1.6
   	release_0_82: 1.4
   	release_0_81: 1.3
   	release_0_80public: 1.1
   	release_0_80: 1.1
   	release_0_79: 1.1
   	release_0_78: 1.1
   keyword substitution: kv
   total revisions: 6;	selected revisions: 1
   description:
Makefile.in 1.6
   date: 1998/02/22 01:00:56;  author: rwscott;  state: Exp;  lines: +23 -14
   Mostly class structure synchronization, get rid of un-used methods, stub out
   methods that we are supposed to have, get the compress flags right....
   
   A couple of fixes to note.  BulletinBoard was not un-drawing its shadows
   when resized bigger, and the ModifyVerify patch from a few days ago was
   segfaulting on textf/test4.
   
   Now back to the menu system......
   =============================================================================
   
   RCS file: /cvsroot/hungry/lesstif/lib/Xm/Shadow.c,v
   Working file: Shadow.c
   head: 1.10
   branch:
   locks: strict
   access list:
   symbolic names:
   	release_0_83: 1.10
   	release_0_82: 1.8
   	release_0_81: 1.8
   	release_0_80public: 1.4
   	release_0_80: 1.4
   	release_0_79: 1.4
   	release_0_78: 1.3
   keyword substitution: kv
   total revisions: 10;	selected revisions: 1
   description:
Makefile.in 1.10
   date: 1997/11/21 18:01:16;  author: miers;  state: Exp;  lines: +2 -8
   Shadow.c: Highlights are supposed to use XFillRectangle to get the proper
   highlight pixmap effects.
   
   DrawingArea.c, Gadget.c, Manager.c, Primitive.c, RowColumn.c: semantics
   of widget_navigable weren't quite right.  The exclusivity of the shell
   is only important if the widgets are TAB_NAVIGABLE.
   
   Gadget.c, Primitive.c: When FOCUS_OUT, set HaveTraversal to False, not True
   (focus_change).
   
   VirtKeys.c: patch from the list.
   
   Primitive.c, Manager.c, Visual.c: The function is supposed to be named
   _XmClearBGPixmap(), not _XmClearBGPixmapName().
   
   TearOffB.c: warning removal from the event loop.  I gotta compile on linux
   more often.
   =============================================================================
   
   RCS file: /cvsroot/hungry/lesstif/lib/Xm/ShellExt.c,v
   Working file: ShellExt.c
   head: 1.3
   branch:
   locks: strict
   access list:
   symbolic names:
   	release_0_83: 1.3
   	release_0_82: 1.2
   	release_0_81: 1.2
   	release_0_80public: 1.1
   	release_0_80: 1.1
   	release_0_79: 1.1
   	release_0_78: 1.1
   keyword substitution: kv
   total revisions: 3;	selected revisions: 1
   description:
Makefile.in 1.3
   date: 1998/02/22 01:00:57;  author: rwscott;  state: Exp;  lines: +7 -45
   Mostly class structure synchronization, get rid of un-used methods, stub out
   methods that we are supposed to have, get the compress flags right....
   
   A couple of fixes to note.  BulletinBoard was not un-drawing its shadows
   when resized bigger, and the ModifyVerify patch from a few days ago was
   segfaulting on textf/test4.
   
   Now back to the menu system......
   =============================================================================
   
   RCS file: /cvsroot/hungry/lesstif/lib/Xm/Simple.c,v
   Working file: Simple.c
   head: 1.13
   branch:
   locks: strict
   access list:
   symbolic names:
   	release_0_83: 1.13
   	release_0_82: 1.10
   	release_0_81: 1.7
   	release_0_80public: 1.5
   	release_0_80: 1.5
   	release_0_79: 1.3
   	release_0_78: 1.3
   keyword substitution: kv
   total revisions: 13;	selected revisions: 1
   description:


TearOff.c Current rev 1.24 Last Release 1.23
TearOff.c 1.24
   date: 1998/03/17 19:59:32;  author: u27113;  state: Exp;  lines: +22 -6
   Fixes by Andreas Zeller


Text.c Current rev 1.54 Last Release 1.50
Text.c 1.54
   date: 1998/04/16 19:19:48;  author: u27113;  state: Exp;  lines: +2 -6
   Text.c : back out change to XmTextSetString. This should fix complaints
   	about it changing the position shown when it shouldn't.
   
   Others: as indicated by Alexander Mai <mai@migdal.ikp.physik.tu-darmstadt.de>,
   	#include <LTconfig.h> should precede (almost) all other includes
   	in the sources.
Text.c 1.53
   date: 1998/04/11 14:52:32;  author: u27113;  state: Exp;  lines: +2 -109
   lib/Xm/ResConvert.c
   	A better implementation that was submitted by Alex Lukyanov a long
   	time ago, but overlooked.
   lib/Xm/Text.c
   	Fix nedit's status line: XmTextSetString and XmTextReplace should
   	show the last position that they modified.
Text.c 1.52
   date: 1998/04/11 10:09:11;  author: u27113;  state: Exp;  lines: +10 -2
   Man pages : include LGPL text so they are licensed by it too.
   
   Text.c : XmTextSetString and XmTextReplace should leave the widget
   	in a state showing the last text that was modified.
   	Should fix nedit's statistics line.
Text.c 1.51
   date: 1998/03/28 23:28:23;  author: u27113;  state: Exp;  lines: +99 -2
   Text.c, TextF.c :
   	Merge traits stuff into lib/Xm (from lib/Xm-2.0),
   	it's not worth it keeping two sources for just this.
   	Makefile.am's corrected for this; the old sources (in Xm-2.0)
   	aren't removed yet though.
   
   Text.c :
   	Also modified _XmTextUpdateLineTable so it behaves well in
   	case we tell it to scan the whole text. It had a tendency to
   	forget putting a terminator behind the last line, which caused
   	stuff to be drawn more than once. Xinvest's help showed this
   	after clicking on a few different topics.
   	Should be fixed now (by a - rather harmless - hack though).
   
   TextStrSo.c : initialize the StringSource to an empty string if
   	we're in a XmTextSetString(w, "");


TextF.c Current rev 1.54 Last Release 1.39
TextF.c 1.54
   date: 1998/05/14 22:57:59;  author: u27113;  state: Exp;  lines: +14 -11
   Makefile.am's : replace @prefix@ by $(prefix) to make this an installation
   	time option rather than a configure time option.
   
   lib/Xm/TextF.c : Try to get XmClipboard API to work here.
   	Grrr. I've been tracking a typo for hours, but now it does
   	at least something more or less right.
   	All this is still #ifdeffed out.
   lib/Xm/CutPaste.c : add XdbDebugs, one NULL test.
TextF.c 1.53
   date: 1998/05/13 21:06:29;  author: u27113;  state: Exp;  lines: +62 -9
   lib/Xm/PanedW.c, lib/Xm-2.0/PanedW.c
   	Jon A. Christopher's fix.
   
   lib/Xm/TextF.c
   	Some work for using XmClipboard API (incomplete)
   
   lib/Xm/CutPaste.c
   	A couple of XdbDebugs + one core dump prevention
   
   CURRENT_NOTES
   	Comment added about using XmClipboard API in XmTextField.
TextF.c 1.52
   date: 1998/05/07 22:23:08;  author: u27113;  state: Exp;  lines: +26 -3
   Adapted the build so it not only figures out whether we have a
   gethostname, but also whether it is in -lnsl or not.
   
   According to Michael Vanier <mvanier@bbb.caltech.edu> we need to
   link in -lnsl on Solaris 2.3 when building mwm in order to get
   gethostname().
   
   This fix should solve that.
   
   TextF.c : add a couple of stubs, nothing real yet, to call XmClipboard*().
TextF.c 1.51
   date: 1998/05/02 14:13:16;  author: u27113;  state: Exp;  lines: +119 -5
   Working.html : add JDK (Java Development Kit). The PowerPC Linux version
   	releases linked with LessTif.
   TextF.c : first cut at implementing Shift-Insert, Ctrl-Insert
   	which should work with the so-called clipboard.
   RowColumnP.h : remove an inconsistency that I introduced (so it seems)
   	when adding render table stuff.
TextF.c 1.50
   date: 1998/04/09 12:23:47;  author: u27113;  state: Exp;  lines: +7 -33
   lib/Xm/GadgetUtil.c : Rename _XmInputInGadget for 2.0 for Mozilla.
   lib/Xm/ScrolledW.c : Don't ask for new geometry if not realized.
   	Maybe this fixes Alexander Mai's XACC problem.
   lib/Xm/TextF.c : Clean up some junk.
   lib/Xm/misc.c : Mozilla build script looks for an identifying string
   	in the shared library (in xmversion.sh).
   doc/lessdox/Makefile.am : add subdirectory widgets
TextF.c 1.49
   date: 1998/04/07 20:38:19;  author: u27113;  state: Exp;  lines: +4 -3
   Vendor.c : fix for 64-bit systems from Alexander V. Lukyanov <lav@yars.free.net>
   ResConvert.c : Add a converter for Xinvest.
   XmStrDefs.h : add XmNlist for Mozilla (only if Motif 2.0).
   lib/Xm/TextF.c : fix by Kevin B. Hendricks <kbhend@dogwood.tyler.wm.edu>
TextF.c 1.48
   date: 1998/04/02 21:43:54;  author: rwscott;  state: Exp;  lines: +16 -3
   implement the synthetic resource export proc for value
TextF.c 1.47
   date: 1998/03/31 03:39:32;  author: rwscott;  state: Exp;  lines: +14 -5
   expose() - Xbae examples/matrix was throwing TextF into an infinite loop of
   	exposes.  Since we update the entire display on each expose event, if
   	there are pending exposes just return, we will update on the next one.
   	Also DrawAll() does a CursorDraw() which expose was doing just after
   	the DrawAll().  This is now gone.
   
   Xbae examples/matrix is now usable with an expose at about the blink rate when
   it has focus. What is Xbae doing with TextF that causes this??  There are
   TextF's all over the place that do not show this!
TextF.c 1.46
   date: 1998/03/28 23:28:19;  author: u27113;  state: Exp;  lines: +80 -4
   Text.c, TextF.c :
   	Merge traits stuff into lib/Xm (from lib/Xm-2.0),
   	it's not worth it keeping two sources for just this.
   	Makefile.am's corrected for this; the old sources (in Xm-2.0)
   	aren't removed yet though.
   
   Text.c :
   	Also modified _XmTextUpdateLineTable so it behaves well in
   	case we tell it to scan the whole text. It had a tendency to
   	forget putting a terminator behind the last line, which caused
   	stuff to be drawn more than once. Xinvest's help showed this
   	after clicking on a few different topics.
   	Should be fixed now (by a - rather harmless - hack though).
   
   TextStrSo.c : initialize the StringSource to an empty string if
   	we're in a XmTextSetString(w, "");
TextF.c 1.45
   date: 1998/03/27 11:23:40;  author: rwscott;  state: Exp;  lines: +10 -3
   set_values() - We cannot call ChangeGC if we are not realized.  This is not
   	a problem since the GC's will be created when we do get realized.
TextF.c 1.44
   date: 1998/03/26 23:28:53;  author: u27113;  state: Exp;  lines: +6 -10
   Small correction; I copy/pasted the wrong couple of lines.
   Now we do update XtHeight in set_values; unfortunately that
   doesn't seem to have any effect :-(
TextF.c 1.43
   date: 1998/03/25 21:09:51;  author: u27113;  state: Exp;  lines: +75 -12
   FontList.c : remove some memory leaks;
   	more importantly: don't return our argument when NULL passed
   	in XmFontListAppendEntry; this doesn't work with the memory
   	allocation rules.
   
   TextF.c (both) : watch out more carefully about when you XmFontListFree.
   	Yesterday's implementation did one where it shouldn't.
   	Also do whatever's necessary when changing the XmNfontList
   	through SetValues. Now the newly adapted test/textf/test2.c works.
   
   test/textf/test2.c : call setvalues on XmNfontList to show what didn't work.
   
   ScrolledW.c: add some XdbDebugs, also strip a couple of bits off what
   	GeometryManager returns.
TextF.c 1.42
   date: 1998/03/25 04:34:24;  author: rwscott;  state: Exp;  lines: +6 -3
   _XmTextFieldFocusOut() - If the widget is not managed do not call the
   	losing focus callback.
   
   This is what was causing the segfault in mgv when return was pressed in the
   magstep text field. Since the activate and losingfocus callbacks called the
   same proc. Activate would call the proc which would unmanage the textfield
   which would trigger the losingFocus callback which would call the same
   proc and start all over again.
TextF.c 1.41
   date: 1998/03/23 20:12:40;  author: u27113;  state: Exp;  lines: +15 -2
   Deal with changed fontlist in SetValues by copying it.
   This should fix Uncle George <gatgul@voicenet.com>'s problem.
   
   Not good enough to change the font internally though.
TextF.c 1.40
   date: 1998/03/15 18:44:11;  author: rwscott;  state: Exp;  lines: +13 -11
   patch from Andreas Zeller for the annoying beeps from text when a modifier
   is pressed


TextIn.c Current rev 1.28 Last Release 1.26
TextIn.c 1.28
   date: 1998/04/16 21:46:45;  author: u27113;  state: Exp;  lines: +5 -2
   TextIn.c : don't allow user to paste into a widget that's not editable.
   	Reported by Calum Grant <Calum.Grant@cl.cam.ac.uk>.
   VirtKeys.c Primitive.c Manager.c :
   	Try to fix the fact that PageDown doesn't work in nedit.
   	The last update on VirtKeys.c inadvertently already implemented
   	a hack which makes it work.
   
   	It now looks like I found an inelegant but better solution :
   	- nedit shows that PageUp behaves like PageLeft and PageDown
   	  behaves like PageRight.
   	- a look at the virtual keys table shows that these are some
   	  of the keys having a modifier.
   	- the comments in VirtKeys.c (from Harald Albrecht ?) say that
   	  CheckForVirtualKey gets called multiple times for every key.
   	  This is not the case: this only happens the first time the user
   	  hits a key in an application. This makes me think that Xt
   	  caches some of this. Calling XtSetKeyTranslator seems to
   	  flush the cache, so it must be Xt, not Xlib.
   	- There already was an _XmVirtualKeysHandler which puts a copy
   	  of the last key pressed in the XmDisplay widget. It was never
   	  called in all of LessTif so I added it to all widgets by
   	  putting one extra line in Primitive.c and Manager.c.
   	  I don't suppose I need to do this for Gadgets as they don't
   	  get events anyway. Manager will hopefully do that for me.
   	- The info from the event (from _XmVirtualKeysHandler)
   	  contains info about which modifiers are in effect.
   	  I'm passing this instead of the usual modifiers to CheckForVirtualKey
   	  in XmTranslateKey. Not to XtTranslateKey.
   	  Not 100ure that this is right but it seems to work.
   
   Be assured that I left out all the hours of testing, digging, trying
   which led nowhere before I figured this out.
   
   In Dutch we have a term "bit-neuken" which translates quite accurately
   into bit-fucking. This is what I had to do to find this. Sigh. Poor bits,
   poor me.
TextIn.c 1.27
   date: 1998/03/15 18:44:12;  author: rwscott;  state: Exp;  lines: +8 -8
   patch from Andreas Zeller for the annoying beeps from text when a modifier
   is pressed


TextOut.c Current rev 1.42 Last Release 1.41
TextOut.c 1.42
   date: 1998/05/06 21:32:11;  author: u27113;  state: Exp;  lines: +11 -2
   Fix mwm XPM icons.
   TextOut.c : implement the hack by vogt@kvi.nl.


TextStrSo.c Current rev 1.17 Last Release 1.15
TextStrSo.c 1.17
   date: 1998/05/02 21:01:08;  author: rwscott;  state: Exp;  lines: +12 -2
   TextStrSo.c:GetSelection() - In ml the Insert Text button in the Reply
       window calls XmTextGetSelectionPosition and ignores the return value.
       Instead it compares the value of left and right to determine if
       anything is selected.  Therefore always setting left and right makes
       it happy.  If we do not have the selection left == right anyway, so
       everyone should be happy and left and right are always valid this way.
TextStrSo.c 1.16
   date: 1998/03/28 23:28:24;  author: u27113;  state: Exp;  lines: +11 -2
   Text.c, TextF.c :
   	Merge traits stuff into lib/Xm (from lib/Xm-2.0),
   	it's not worth it keeping two sources for just this.
   	Makefile.am's corrected for this; the old sources (in Xm-2.0)
   	aren't removed yet though.
   
   Text.c :
   	Also modified _XmTextUpdateLineTable so it behaves well in
   	case we tell it to scan the whole text. It had a tendency to
   	forget putting a terminator behind the last line, which caused
   	stuff to be drawn more than once. Xinvest's help showed this
   	after clicking on a few different topics.
   	Should be fixed now (by a - rather harmless - hack though).
   
   TextStrSo.c : initialize the StringSource to an empty string if
   	we're in a XmTextSetString(w, "");


ToggleB.c Current rev 1.29 Last Release 1.28
ToggleB.c 1.29
   date: 1998/03/16 21:52:18;  author: rwscott;  state: Exp;  lines: +12 -15
   I could not get Andrea's varargs patch to go in clean so I tried it by hand.
   
   defined the rcsid's as const
   
   cleaned out some of the unused variables.


ToggleBG.c Current rev 1.28 Last Release 1.26
ToggleBG.c 1.28
   date: 1998/03/18 20:33:56;  author: rwscott;  state: Exp;  lines: +4 -4
   _XmToggleButtonGadgetSetState && XmToggleButtonGadgetSetState - from
   	Michel Bardiaux <mbardiaux@usrconsult.be>
   	The SetState code in ToggleB.c is almost the same as in ToggleBG.c, but
   it does not have that trouble because it has its own window, which is
   unmapped, so any draws just end in the bit bucket. ToggleBG, however,
   borrows a window from its parent, and *that* is mapped even though the
   gadget is not.
ToggleBG.c 1.27
   date: 1998/03/16 21:52:27;  author: rwscott;  state: Exp;  lines: +12 -14
   I could not get Andrea's varargs patch to go in clean so I tried it by hand.
   
   defined the rcsid's as const
   
   cleaned out some of the unused variables.


TrackLoc.c Current rev 1.3 Last Release 1.2
TrackLoc.c 1.3
   date: 1998/03/19 20:07:04;  author: u27113;  state: Exp;  lines: +23 -12
   FileSB.c: something in Eric Howe's patch drove xmgr into an
   	infinite loop. Fixed that.
   TrackLoc.c : Andreas Zeller's patch.


Transltns.c Current rev 1.5 Last Release 1.4
Transltns.c 1.5
   date: 1998/04/21 17:13:25;  author: u27113;  state: Exp;  lines: +12 -3
   Fix to keep the JDK people happy.
   See mails from Kevin B. Hendricks <kbhend@dogwood.tyler.wm.edu>


Vendor.c Current rev 1.38 Last Release 1.33
Vendor.c 1.38
   date: 1998/04/07 20:38:18;  author: u27113;  state: Exp;  lines: +18 -5
   Vendor.c : fix for 64-bit systems from Alexander V. Lukyanov <lav@yars.free.net>
   ResConvert.c : Add a converter for Xinvest.
   XmStrDefs.h : add XmNlist for Mozilla (only if Motif 2.0).
   lib/Xm/TextF.c : fix by Kevin B. Hendricks <kbhend@dogwood.tyler.wm.edu>
Vendor.c 1.37
   date: 1998/04/01 20:30:44;  author: u27113;  state: Exp;  lines: +2 -4
   Also activate Editres when LESSTIF_PRODUCTION.
Vendor.c 1.36
   date: 1998/03/28 02:34:24;  author: rwscott;  state: Exp;  lines: +4 -2
   Damn, I didn't mean to commit this.  #if 0 out the stuff that I am playing
   with. (Related to setting the transient hint on windows that are realized
   before the window that they are transient for)
Vendor.c 1.35
   date: 1998/03/28 02:24:55;  author: rwscott;  state: Exp;  lines: +26 -2
   _XmFixOptionMenu() - While trying to assign the MemWidget to the first item
   	in the menu, we use the second item if the first is a tear off
   	control.  This causes a problem if the tear off control is the only
   	item in the menu.  Therefore check for at least two items before
   	setting the memWidget.  This shows up now because the tear-off
   	control is being created in initialize instead of when the menu
   	is posted.
   
   
   This is for the one that Jon just reported.
Vendor.c 1.34
   date: 1998/03/18 21:55:12;  author: u27113;  state: Exp;  lines: +4 -4
   Vendor.c : Fix by Simon Gerraty <sjg@quick.com.au> to make Java AWT apps
   	not move to the upper left of the screen when coordinates given
   	are 0,0; it seems they pass 1,1 for that to happen.
   
   MainW.c : One of Alexander Mai's <mai@migdal.ikp.physik.tu-darmstadt.de>
   	complaints: now nedit shows the statistics line again. Somehow
   	XmMainWindow was losing some information that it did have.


VirtKeys.c Current rev 1.8 Last Release 1.6
VirtKeys.c 1.8
   date: 1998/04/16 21:46:41;  author: u27113;  state: Exp;  lines: +59 -16
   TextIn.c : don't allow user to paste into a widget that's not editable.
   	Reported by Calum Grant <Calum.Grant@cl.cam.ac.uk>.
   VirtKeys.c Primitive.c Manager.c :
   	Try to fix the fact that PageDown doesn't work in nedit.
   	The last update on VirtKeys.c inadvertently already implemented
   	a hack which makes it work.
   
   	It now looks like I found an inelegant but better solution :
   	- nedit shows that PageUp behaves like PageLeft and PageDown
   	  behaves like PageRight.
   	- a look at the virtual keys table shows that these are some
   	  of the keys having a modifier.
   	- the comments in VirtKeys.c (from Harald Albrecht ?) say that
   	  CheckForVirtualKey gets called multiple times for every key.
   	  This is not the case: this only happens the first time the user
   	  hits a key in an application. This makes me think that Xt
   	  caches some of this. Calling XtSetKeyTranslator seems to
   	  flush the cache, so it must be Xt, not Xlib.
   	- There already was an _XmVirtualKeysHandler which puts a copy
   	  of the last key pressed in the XmDisplay widget. It was never
   	  called in all of LessTif so I added it to all widgets by
   	  putting one extra line in Primitive.c and Manager.c.
   	  I don't suppose I need to do this for Gadgets as they don't
   	  get events anyway. Manager will hopefully do that for me.
   	- The info from the event (from _XmVirtualKeysHandler)
   	  contains info about which modifiers are in effect.
   	  I'm passing this instead of the usual modifiers to CheckForVirtualKey
   	  in XmTranslateKey. Not to XtTranslateKey.
   	  Not 100ure that this is right but it seems to work.
   
   Be assured that I left out all the hours of testing, digging, trying
   which led nowhere before I figured this out.
   
   In Dutch we have a term "bit-neuken" which translates quite accurately
   into bit-fucking. This is what I had to do to find this. Sigh. Poor bits,
   poor me.
VirtKeys.c 1.7
   date: 1998/04/16 19:19:45;  author: u27113;  state: Exp;  lines: +31 -5
   Text.c : back out change to XmTextSetString. This should fix complaints
   	about it changing the position shown when it shouldn't.
   
   Others: as indicated by Alexander Mai <mai@migdal.ikp.physik.tu-darmstadt.de>,
   	#include <LTconfig.h> should precede (almost) all other includes
   	in the sources.


Visual.c Current rev 1.10 Last Release 1.9
Visual.c 1.10
   date: 1998/03/24 23:01:13;  author: rwscott;  state: Exp;  lines: +112 -77
   From:        Eric Howe <mu@wintermute.echo-on.net>
   To:          lesstif@Hungry.COM
   Subject:     Colormap patch
   Date:        Tue, 24 Mar 1998 03:32:40 -0500 (EST)
   
   
   I think I've fixed up LessTif to do the "right thing" when an application
   is using non-default colormaps.  I've only tested the patch with mgv
   (run mgv with the -install switch to get a private colormap) but there
   shouldn't be any problems with other Motif applications.  The patch was
   made against lesstif-current from 1998.03.23.
   
   A summary of the changes follows:
   
   	include/Motif-1.2/XmI/MacrosI.h
   		Added the ColormapOfObject macro.
   
   	include/Motif-1.2/XmI/XmI.h
   		Added prototypes for the new _XmWhitePixelOfObject and
   		_XmBlackPixelOfObject functions (the implementations
   		are in Visual.c).
   
   	lib/Xm/ArrowB.c	lib/Xm/ArrowBG.c
   	lib/Xm/Label.c	lib/Xm/LabelG.c
   	lib/Xm/List.c
   		WhitePixelOfScreen and BlackPixelOfScreen changed to
   		_XmWhitePixelOfObject and _XmBlackPixelOfObject.
   
   	lib/Xm/MainW.c
   	lib/Xm/ScrollBar.c
   	lib/Xm/ScrolledW.c
   		WhitePixelOfScreen and BlackPixelOfScreen changed to
   		_XmWhitePixelOfObject and _XmBlackPixelOfObject;
   		DefaultColormapOfScreen changed to CoreColormap.
   
   	lib/Xm/Visual.c
   		WhitePixelOfScreen and BlackPixelOfScreen changed to
   		_XmWhitePixelOfObject and _XmBlackPixelOfObject;
   		DefaultColormapOfScreen changed to CoreColormap.
   		Added the _XmWhitePixel, _XmBlackPixel,
   		_XmWhitePixelOfObject, and _XmBlackPixelOfObject
   		functions.  _XmWhitePixel and _XmBlackPixel had
   		prototypes in Xm/XmP.h but no implementations,
   		_XmWhitePixelOfObject and _XmBlackPixelOfObject
   		are more convenient forms of _XmWhitePixel and
   		_XmBlackPixel.
   
   Since the drag-and-drop code draws on the root window (AFAIK), I left it
   using the default colormap.
   
   A (very) short summary of what to do when you need a Colormap or black/white
   Pixel value is at
   
   	http://www.echo-on.net/~mu/lesstif-colors.html


XmOS2.def.dist Current rev 1.2 Last Release 1.1
XmOS2.def.dist 1.2
   date: 1998/04/07 18:18:44;  author: u27113;  state: Exp;  lines: +1793 -1723
   Patch sent by Michael Holzapfel <michael@ime.rwth-aachen.de>


XmString.c Current rev 1.19 Last Release 1.16
XmString.c 1.19
   date: 1998/04/16 19:19:46;  author: u27113;  state: Exp;  lines: +4 -85
   Text.c : back out change to XmTextSetString. This should fix complaints
   	about it changing the position shown when it shouldn't.
   
   Others: as indicated by Alexander Mai <mai@migdal.ikp.physik.tu-darmstadt.de>,
   	#include <LTconfig.h> should precede (almost) all other includes
   	in the sources.
XmString.c 1.18
   date: 1998/04/14 23:13:03;  author: rwscott;  state: Exp;  lines: +10 -6
   Whoa, missed this.  Wrap the fprintf's of the string stats in XdbDebug's.
   
   Peter, to get your xm string stats to show like you intended run the apps
   with the environment variable DEBUGSOURCES=XmString.c.
XmString.c 1.17
   date: 1998/04/14 22:54:50;  author: rwscott;  state: Exp;  lines: +80 -2
   FileSB.c LabelG.c SelectionBox.c XmString.c Xmos.c
   
   Patch from Peter Wainwright <prw@wainpr.demon.co.uk> dealing mainly with
   memory leaks during string handling.
   
   List.c:_XmListSetGeometry() - If the parent is not managed we were not asking
   	for a geo change.  The result was that a change of visibleItems while
   	the parent was not managed was not changing the visible items.  If the
   	parent is not managed Xt should take care of the size request by
   	just granting it.  This was messing up one of my apps.  If someone
   	can point out why it was there I will re-evaluate this.
   
   BulletinBoard.c:handle_change_managed() - If we are not realized, layout with
   	the current width/height, this will honor the specified resources if
   	any.


XmXpm.c Current rev 1.3 Last Release 1.1
XmXpm.c 1.3
   date: 1997/08/19 19:29:26;  author: miers;  state: dead;  lines: +0 -0
   More code reorg.
   
   1) Moved all the Xme functions to the 2.0 tree.  Something needs to be
   done about the headers, though.
   
   2) If LESSTIF_PRODUCTION is defined, then the Editres handler and the
      pixmap extensions aren't built.  I'd like to move these to another
      place (e.g., extensions/) but it's so much trouble to hack configure
      that I'm putting it off.
   
   3) Add an Editres for R5.
XmXpm.c 1.2
   date: 1997/07/12 18:37:10;  author: miers;  state: Exp;  lines: +1242 -862
   Indenting done -- whew.


Xmos.c Current rev 1.17 Last Release 1.12
Xmos.c 1.17
   date: 1998/04/16 19:19:47;  author: u27113;  state: Exp;  lines: +4 -3
   Text.c : back out change to XmTextSetString. This should fix complaints
   	about it changing the position shown when it shouldn't.
   
   Others: as indicated by Alexander Mai <mai@migdal.ikp.physik.tu-darmstadt.de>,
   	#include <LTconfig.h> should precede (almost) all other includes
   	in the sources.
Xmos.c 1.16
   date: 1998/04/15 00:20:05;  author: rwscott;  state: Exp;  lines: +4 -3
   SelectionBox.c:initialize() - We were calling set_values here to set the
   	default button.  This meant calling set_values on a widget that is
   	not yet fully initialized.  This was probably not a big deal for
   	SelectionBox but, sub-classes have not even had there initialize
   	method called yet!  In the case of FSB the set_values was being
   	called before the initialize, which meant most of the FSB had not
   	been built yet!  Use the BB_ macro to set the default button since
   	it is a sub-class of BB anyway.
   
   This fixes the recent core dumps in the filesb tests. This also seems to fix
   the "no shadow on the menubars" in some of the FSB tests also.
   
   Xmos.c:_XmOSBuildFileList() - wrap another fprintf in an XdbDebug
Xmos.c 1.15
   date: 1998/04/14 22:54:51;  author: rwscott;  state: Exp;  lines: +7 -2
   FileSB.c LabelG.c SelectionBox.c XmString.c Xmos.c
   
   Patch from Peter Wainwright <prw@wainpr.demon.co.uk> dealing mainly with
   memory leaks during string handling.
   
   List.c:_XmListSetGeometry() - If the parent is not managed we were not asking
   	for a geo change.  The result was that a change of visibleItems while
   	the parent was not managed was not changing the visible items.  If the
   	parent is not managed Xt should take care of the size request by
   	just granting it.  This was messing up one of my apps.  If someone
   	can point out why it was there I will re-evaluate this.
   
   BulletinBoard.c:handle_change_managed() - If we are not realized, layout with
   	the current width/height, this will honor the specified resources if
   	any.
Xmos.c 1.14
   date: 1998/04/08 00:12:11;  author: rwscott;  state: Exp;  lines: +25 -2
   FileSB.c:defaultQualifySearchDataProc() - Completely re-write this puppy.
   	Use filesb/test9 to check it out.
   FileSB.c:_XmFsbButton() - Use the Qualify proc to do the filtering.
   
   Xmos.c:_XmOSQualifyFileSpec() - Never return a relative path directory. If
   	it comes in relative tack on the cwd.
   
   Beware, I still have some questions about this.  I have not tried it with
   Mozilla yet but, it seems to keep xmgr happy even with the strange things
   it is doing.
Xmos.c 1.13
   date: 1998/04/04 03:07:07;  author: rwscott;  state: Exp;  lines: +6 -2
   Related to segfaults and infinite loops with file selection in Mozilla.
   This gets the dialog to show, but traversing the file system is another thing.
   It uses its own qualify/dir/file search procs. That do not seem to be all
   that compatible with how we do things.  Now to locate them in the Mozilla
   source to see what is going on!


misc.c Current rev 1.7 Last Release 1.6
misc.c 1.7
   date: 1998/04/09 12:23:49;  author: u27113;  state: Exp;  lines: +3 -1
   lib/Xm/GadgetUtil.c : Rename _XmInputInGadget for 2.0 for Mozilla.
   lib/Xm/ScrolledW.c : Don't ask for new geometry if not realized.
   	Maybe this fixes Alexander Mai's XACC problem.
   lib/Xm/TextF.c : Clean up some junk.
   lib/Xm/misc.c : Mozilla build script looks for an identifying string
   	in the shared library (in xmversion.sh).
   doc/lessdox/Makefile.am : add subdirectory widgets


Executive Summary
Of 73 files, 41 changed, and 32 had no change

File            Current Rev. Last Release
.indent.pro     1.1          1.1    No Change
ArrowB.c        1.18         1.17   
ArrowBG.c       1.13         1.12   
AtomMgr.c       1.3          1.3    No Change
BaseClass.c     1.9          1.9    No Change
BulletinBoard.c 1.24         1.21   
Cache.c         1.7          1.7    No Change
CascadeB.c      1.82         1.80   
CascadeBG.c     1.75         1.74   
Command.c       1.9          1.9    No Change
CutPaste.c      1.6          1.4    
DebugUtil.c     1.26         1.25   
DefaultProcs.c  1.9          1.5    
DragIcon.c      1.7          1.7    No Change
DragOverS.c     1.8          1.8    No Change
DragUnder.c     1.2          1.2    No Change
DrawingArea.c   1.21         1.20   
DrawnB.c        1.23         1.23   No Change
DropSMgr.c      1.15         1.15   No Change
DropTrans.c     1.8          1.8    No Change
ExtObject.c     1.3          1.3    No Change
FileSB.c        1.38         1.24   
FontList.c      1.13         1.10   
Form.c          1.109        1.106  
Frame.c         1.35         1.34   
Gadget.c        1.11         1.11   No Change
GadgetUtil.c    1.7          1.6    
GeoUtils.c      1.22         1.18   
Hash.c          1.3          1.2    
ImageCache.c    1.9          1.9    No Change
LTCvt.c         1.4          1.4    No Change
LTEditres.c     1.6          1.1    
LTV5Editres.c   1.2          1.2    No Change
LTV6Editres.c   1.3          1.2    
LTXpm.c         1.3          1.3    No Change
LTfnmatch.c     1.2          1.2    No Change
LTglob.c        1.3          1.3    No Change
Label.c         1.63         1.58   
LabelG.c        1.54         1.51   
List.c          1.50         1.38   
MainW.c         1.28         1.26   
Makefile.am     1.9          1.9    No Change
Makefile.in     1.15         1.13   
TearOff.c       1.24         1.23   
TearOffB.c      1.14         1.14   No Change
Text.c          1.54         1.50   
TextF.c         1.54         1.39   
TextIn.c        1.28         1.26   
TextOut.c       1.42         1.41   
TextStrSo.c     1.17         1.15   
ToggleB.c       1.29         1.28   
ToggleBG.c      1.28         1.26   
TrackLoc.c      1.3          1.2    
Transltns.c     1.5          1.4    
TravAct.c       1.7          1.7    No Change
Traversal.c     1.23         1.23   No Change
UniqueEvent.c   1.3          1.3    No Change
VaSimple.c      1.2          1.2    No Change
Vendor.c        1.38         1.33   
VirtKeys.c      1.8          1.6    
Visual.c        1.10         1.9    
World.c         1.3          1.3    No Change
XmIm.c          1.4          1.4    No Change
XmOS2.def.dist  1.2          1.1    
XmOS2.def.orig  1.2          1.2    No Change
XmStrDefs.c     1.2          1.2    No Change
XmString.c      1.19         1.16   
XmXpm.c         1.3          1.1    
Xmos.c          1.17         1.12   
alloca.c        1.2          1.2    No Change
inlines.in      1.3          1.3    No Change
misc.c          1.7          1.6    
mkinline.c      1.3          1.3    No Change
