Known bugs in the original Fermilab Histo-Scope v4.0
----------------------------------------------------

1. hs_title simply overwrites the previous title string instead of 
   doing free/strdup. As a result the program memory becomes corrupted
   if the new title string is longer than the old one.

2. Histo-Scope plots are not updated by hs_set_1d_errors and hs_set_2d_errors.

3. Positive and negative errors are swapped in 1-d histogram plots.

4. Positive and negative horizontal errors are swapped in "XY plots
   with errors" produced by the ntuple panel.

5. There is a strange hs_load_config_string and hs_load_config_file bug:
   calling these commands more than 3 times causes core dump in the
   connected Histo-Scopes. This bug strongly limits your ability to build
   your own dynamic interface for displaying histogram windows.

6. "tohbook" program does not convert ntuples correctly

7. hs_read_file creates a bad category name (which ends with slash) for
   uncategorized items when the category prefix is not an empty string.
   Such category name can not be correctly parsed by Histo-Scope.

8. hs_read_file creates a bad category name (which starts with slash) for
   items in any category when the category prefix argument is an empty string.
   Such category name can not be correctly parsed by Histo-Scope.

9. Multiple requests to load configuration strings which display overlayed
   plots cause Histo-Scope core dump because of an attempt to free some 
   memory more than once.

10. There is a bug in the histogram updating code which causes the program
    to disconnect from a client when the size of the update request message
    from the client is large enough so that this message can not be received
    in a single read from a socket. This bug prevents you from displaying
    a large number of histograms simultaneously.

11. There is a bug in the code which updates Histo-Scope "Windows" menu.
    Depending on the state Histo-Scope is in, this bug may cause core
    dumps during window postings/removals. It may actually be the same
    bug as the one described in item 5.

12. There is a bug (feature?) in the Histo-Scope code which receives
    configuration strings. Histo-Scope writes the string into a temporary
    file first and then loads it as a configuration file. However, the
    temporary file is not deleted aftewards. This may result in creation
    of a huge number of /tmp/histo* files.

13. An attempt to create an adaptive 1d histogram from an empty ntuple
    crashes Histo-Scope.

14. Histo-Scope plots are not updated by the hs_hist_set_gauss_errors
    function. 

15. hs_update produces rather annoying messages "Ignoring request for
    nonexistant item from HistoScope" when currently displayed items
    are deleted.

16. Repetitive clicks on the "delete window" button in the plot windows
    (usually displayed by the window manager as a small cross in the upper
    right corner) crash the Histo-Scope GUI.

17. The following sequence causes the Histo-Scope GUI to emit a wrong
    error message that a histogram has no errors:
      -- Start the GUI.
      -- Create a histogram. Fill it.
      -- Set the errors with something like hs_set_1d_errors.
      -- Load a config file or string which requests the histogram
         to be displayed with errors.
    The problem here is that the info about a histogram is sent to the
    GUI only when the histogram is created (without errors), and it is
    not updated when the errors are first set. 

18. A bug in the scatter plot widget prevents it from displaying color
    scatter plots.

19. A bug in the vertical scale layout code often causes axis ticks to be
    misplaced.

20. The plotting widgets redraw their images for every "expose" event
    from the X server rather than for the last event in a sequence of
    similar events. With some window managers, this results in lots of
    unnecessary redrawing operations whenever the windows are moved or
    resized.

21. The code which renders text in PostScript fails if a string contains
    backslashes or unbalanced parentheses.

22. A histogram created from the ntuple interface has a wrong value in its
    last bin.

23. An attempt to remove an overlayed plot whose item has been deleted
    crashes the Histo-Scope GUI.

24. Some pixmaps are not properly deallocated on the X server by the line,
    mark, and fill style selection dialogs.

25. Histo-Scope GUI disconnects from a client spuriously when it displays
    an ntuple plot while the ntuple is being reset by the client. This
    happens under following circumstances:
      -- Histo-Scope requests the rest of the data for a partially
         transferred ntuple.
      -- The client resets the ntuple (a message is sent to Histo-Scope).
      -- The client refills the ntuple.
      -- The client answers the Histo-Scope request, treating the ntuple
         as if it was already partially transferred.
    As a result, Histo-Scope misses the first part of the ntuple data.
    It generates an error message and disconnects.

26. "View", "View Overlaid", and "View Multiple" Histo-Scope GUI buttons
    do not work on big-endian platforms.

27. Function "redisplaySubcategories" in file mainPanel.c has an unchecked
    array bound which results in a buffer overrun if the number of
    categories to show in the GUI window is larger than 41.