	VNC 42.75

- I'm using now OpenWindowTagList() instead of
  OpenWindow(). 

- ViNCEd does now no longer insert the dummy
  gadget into the window border if it really doesn't
  have to. This might help to work around the broken
  MCP window border hack. This does not mean that
  this problem won't bite back sometimes. There are,
  in fact, situations where this gadget is required, for
  example if "NOSIZE" is selected. 

- I replaced the default iconification image by
  something which looks more what people seem to
  prefer. This is now also a pure Boopsi class
  and a look-alike to Massimos Boopsi class.

- With kind permission of Massimo, I include 
  the toolbutton image class again. VNC has now
  its own iconification image, but it is recommended
  to replace it by this class to make the iconify
  image public to the system. Thanks Massimo!

- Updated the NDK to include the toolbutton 
  image includes and autodocs.

- Enhanced the double TAB handling again (oh no,
  not yet another feature). First, the double TAB
  delay time is now selectable and is under control
  of the timing page. Second, if this time is set to
  zero, the TAB requester will always pop up after the
  second TAB.

- Made the default settings aware of this setup.
  Hence, the requester will pop up after the second TAB.
  I hope this will *finally* end this discussion.

- ViNCEd still allocates two boopsi images per arrow.
  The reason is that the sysiclass boopsi comes with
  a specific tag item to select the raised/recessed state.
  I read this as an indicator that not all implementations
  of the sysiclass can handle raised/recessed states 
  automatically. The V40 sysiclass should do however. 
  Since this causes no additional drawback, I will keep 
  it like this.


	VNC 42.76

- ^J (insert CSI) did not work very well with the TAB
  expansion, due to two bugs or oversights in the code.
  Should be fixed now, even though the fix was quite tough.

- The "KeyCursorNRight" primitive did not work as expected
  in case line break at the right window border was turned
  on and the flag "don't scroll into the border" was turned
  on simultaneously. This was a new bug introduced in 42.74.

- The window open code is now about 100% faster simply by
  not sending a reset "ESCc" to the console.device.

- Changed the menu item "Hide" to "Unmark". 

- Added again new features. Feature number one is that
  the TAB expansion can be told now to ignore the data
  behind the cursor in the sense that this data will not
  contribute to the pattern. This improves the usefulness
  of the TAB expansion by about 1% and required one 
  additional TAB expansion flag and one additional TAB
  control page of SetVNC (there are now seven Shell pages)
  decreasing the useabilty of the configuration by 
  about 20%. It also means I have to re-organize the guide and
  have to build a new locale for the new strings.

- Added another new feature. Due to common request, the TAB
  expansion may now list its results on the console instead
  of using the ASL requester. This can be found on the fourth
  shell page as the other new flag.
  Be warned! This feature was a nightmare to implement and it
  really breaks with various issues of good style. First of
  all, it breaks the input mask of programs which is un-
  desirable. Secondly, it messes with data structures of 
  the main task it shouldn't mess with. It required the intro-
  duction of one additional DOS packet and various long
  testing sessions. 


	VNC 42.77

- Border computation was partially wrong, it did not take the
  border protection zone into account, leading to partial
  border trashing on CSI x.

- Macro button expansion did not keep track of the keyboard
  lock correctly and hence might have locked the output in
  situations where it should have been unlocked.


	VNC 42.78

- The window contents re-arrangement on a window resize was
  partially wrong for some setups. This was due to a register
  trashing in one of the low-level routines. *Outch*

- In case iconficiation is forbidden, the second press on the
  iconification gadget will restore the original window 
  position and size.


	VNC 42.79

- The calculation of the iconification image size wasn't always
  optimal. It now closely reassembles the same style and size as
  the intuition code.


	VNC 42.80

- The default preferences still listed Amiga-H instead of Amiga-U
  for "unmark". Similary, the name of the keyboard function was
  still wrong.

- Fixed the wrong layout of the left border of the build-in
  iconification image. The color choice for the leftmost line of
  the "selected" image was wrong.

- Improved the gagdet layout for bitmaps of depth one, i.e. if only
  two pens are available.

- Removed the old Mwm style iconification image completely. This means
  that no iconification gadget can be build in case allocating the
  new boopsi failed.

- Added a new feature: Double clicking on a word captures the complete
  word in a block. 

- Added a new CSI sequence. CSI ?46h turns on logging. All printed
  data will be written over the serial port at 9600 baud, 8 bit, one
  stop bit. Alternatively, you may capture this data with Sashimi.


	VNC 42.81

- Fixed a crash of the buffer IO module that was caused by a register 
  trash in the IO support routines.

- The VNC requester layout routine removes now a terminating LF in the
  requester body to make the requesters looking a bit more fancy.


	VNC 42.82

- ViNCEd forgot to close the initial screen in case the window layout
  of the initial window failed.

NOTE: If you see crashes and hangs on a window open with "unusual"
dimensions, i.e. on tiny screens, then blame your favourite GUI improver
hack for this. ViNCEd is rather bullet proof in this respect.

	VNC 42.83

- ViNCEd forgot to dispose one of its boopsis correctly on an error
  condition. This did not cause any problems until you tried to flush
  the vnc.library, which then resulted in a warning since VNC could
  not unlink its private Boopsi classes then. Fixed.
- Added a somewhat uneffective workaround against a gadtools bug.
  The gadtools menu creator loops forever if the screen is too
  small or the font is too large. VNC tries now to detect this
  situation and will not call gadtools then. However, as it is unclear
  under which circumstances exactly gadtools will lock up, this
  workaround might not be very effective. It should catch the most
  obvious cases, at least. [David]
- Changed the copyright from 2000 to 2001.

	VNC 42.84

- Fixed all the copyright notices to 2001 now. Really.

	VNC 42.85

- Forgot to bump the library release to 42.84, is now 42.85. Fixed.
- TAB expansion behaviour changed slightly now. The TabHook flushes
  its internal list now as soon as you press a different "Expansion"
  key than the expansion was started with. Old behaivour was to ignore
  the expansion type when cycling thru the list of found expansions.
  [Vit Sindlar]

	VNC 42.86

- VNC supports now a new "keyboard type". If the keyboard name "*" or "?"
  is selected, ViNCEd will update the keyboard type automatically to the
  default system keymap as soon as it gets changed (i.e. by the input
  preferences program). Note that this feature is disabled by default 
  for backwards compatibility. To use it, an explicit "SetKeyBoard "?""
  or the equivalent "echo "*E]42;**"" is required. You may want to add 
  this command to your "Shell-Startup" file. [Vit Sindlar]
- Updated the ViNCEd NDK files, brought them up to date and reworked
  some of the example programs a bit.

	VNC 42.87

- Fixed a bug in the button setup routine that broke the gadget list
  completely if the available space for the buttons was not available.

	VNC 42.88

- Save requesters appear now in inverse colors.
- Updated SetVNC 41.48 to use the new requester flags as well.

	VNC 42.89

- The TAB expansion ignores now a trailing '/' or ':' when sorting
  in found expansions. [Allan]

	VNC 42.90

- The "do not match characters behind cursor" option of the TAB
  expansion always tried to concatenate the found file with the
  remaining string. It now uses a heuristic based on the first
  character whether the string should be concatenated or not.
  [Allan]
