FILE: TESTING		WHAT TO TEST (for vi.iv)

KEY:	* = important bug
	+ = not so important bug
	& = not yet checked
	
	Unmarked items are currently fixed.  They are left in this
	list so that we know to test for it in future versions.
	And, if a bug gets re-introduced, we can see in which versions 
	it worked.
----------------------------------------------------------------
  (1) to subsitute LR for RL & vice-versa -- segmentation fault (&
      screws up the screen).  Worked on TX.  Fixed on CS by inserting
      check for null pointer in fcn convlinebp.
		2.1 works
		3.1 works
		5.1 works

* (2) have search pattern for numbers & upper-case match both RL & LR
      versions. 
	[a] When finds target in secondary language, does not position
	cursor correctly.
	
	[b] When entering the target in RL insert mode, it acts as if
	'rlnu' is set, even if it is not.  It'll echo the search
	string you enter according to the actual status of rlnu, but
	when it searches for matches, it acts as if rlnu was set.
      
		Both [a] & [b] were carefully noted for 3.1.  This
		function also did not work for 2.1.  Presumably it had
		the same behavior, but it was not so carefully checked.

+ (3) search for string containing LR & RL chars.  -- WORKS [8/11]
		2.1 works
		3.1 works (but did not position cursor properly when
		the lang of the 1st char != view lang.  This was
		probably also true for 2.1).
		5.1 works (like ver. 3.1)

  (4) 	{this item deleted}

& (5) non-shef files
		2.1 not checked

+ (6) :set chlk (^x).  Manual is  wrong; need the "=".  "changelang"
  works, but "chlk" doesn't.
		2.1 bug is as described
		3.1 bug is as described

& (7) :set bsRL
		2.1 not checked

  (8) while in LR view & inserting RL text that extends across the
      line (& vice-versa).  Doesn't work even when view = language!
      TX -- works when view = lang.  But when view != lang, when it
      wraps to next line, cursor jumps to lower R corner & stays there
      until ESC is hit, then the line appears correctly.
		2.1 bug is as described
		3.1 works
		5.1 works

  (9)	{This item deleted} 

  (10) v command (move cursor to next char in the file)   -- WORKS [8/11]
		2.1 works
		3.1 works
		5.1 works (also the V command)

& (11) check cursor position in entering/exiting insert mode, given
  different permutations of insert languages.
		2.1 not carefully checked

  (12) displays "ERROR" just before displaying the edit buffer.  True
       whether the cmd is viiv or exxe.  Also displays when do shell
       cmd, the output is printed, & viiv is waiting for the CR to
       continue; it displays the "ERROR" in place of the "Hit Return
       to Continue".  Works on TX
		2.1 bug is as described
		3.1 works	(need -DVMUNIX & "xstr")
		5.1 works

+ (13) sometimes when hit space, cursor moves several columns.  It
  might be linked to the "chlk" problem above.
		2.1 cannot duplicate

  (14) when in RL view & typing search pattern, echoes one of the
  buffer lines on the cmd line.  WORKS [21/11]
		2.1 works
		3.1 works
		5.1 works

+ (15) Check the "I" visual-mode cmd.  WORKS [21/11]
		2.1 works
		3.1 works (but if insert mode != view, starts at 2nd
		column.  This bug probably also was in ver 2.1).

+ (16) when type :q, but haven't saved edits, & are in RL view, the
  screen gets redisplayed from left to right, with the text backwards.
  It works OK in LR view. Same thing on TX  [20/11]
		2.1 bug is as described
		3.1 bug is as described (but when I compile without
		-DVMUNIX & without "xstr", it doesn't do this.  See
		(12) on why I don't do this).

+ (17) When typing in a search pattern in RL mode & try to cancel it
  via ^U, it only erases part of the string & echoes "^@".
		3.1 bug is as described (probably also existed for 2.1).

* (18) Sometimes reverses the direction of strings.  This reversal
  remains even after ^L.  This might be only for lines with both LR &
  RL strings.  Not sure yet what causes this.  [05/01/90]
  
  One cause seems to be starting certain visual command sequences &
  then cancelling them with the ESCAPE key.  I could generate this
  error by typing 'y' or 'd' followed by ESCAPE.	[13/02/90]

  (19) see (23)
  
  (20) If put vi.iv commands in .exrc & then run regular vi, vi will
  complain about the commands (but it doesn't hurt anything).  [05/01/90]

	4.1 fixed by adding use of .exrciv & EXINITIV (see ex.c).

* (20) in ~haim/freddy/title.ps, did ":%s/middle/Xctr/g" & got a
  segmentation fault.  Also in ~haim/tmp/odh.sed.save.  [10/01/90]

* (21) When do "cc", inserts string from another place in the file
  after the cursor.  This string remains even after ^L.  Apparently,
  this does not occur if the line you are replacing is indented.
  [10/01/90]

+ (21) In LR view, if have the an RL block followed by an LR space, &
  "marksl" is set, the spaces are marked as RL chars.  Passing over
  them via the -> key changes their marking to LR, but ^L restores the
  RL marking.
  
& (22) While in RL view, do Q cmd to goto line mode.  Then do "vi" to
  return to visual mode.  See if displays correctly.  If not, does ^L
  restore it?  If not, what happens with ":set lr:"? [10/01/90]

  (23) Not all of my function keys work (maybe because termcap
  definition is based on vi100, not vt100.  [10/01/90]
	
	4.1 fixed by making new termcap definition (vt100iv)

+ (24) Doesn't always handle tabs correctly during insert.  Go to
  indented line (with "autoindent' set), do "O" or "o" cmd.  Do ^D
  once or twice, then start inserting characters.  ^L clears this up. 
  I suspect that this bug did NOT exist before Uri made his fixes.  
  [10/01/90]

+ (25) The "echo" (i.e., the command line at the bottom of the screen)
  sometimes get scrolled up into the text area.  ^L clears this up.
  This probably occurs because Uri had commented out a line in function 
  "selbuf" in file ex_RL.c (because that line caused a worse problem;
  see the code there).		[20/02/90]
  
* (26) When switch to cmd mode (e.g., via Q command), don't get the
  ":" prompt.  This bug might be new.	[04/03/90]
  
	4.1	bug 1st noted

+ (27) Under the following conditions, the line-number field might be
  underlined.  ^L will clear it.
	[a] RL view
	[b] insert mode
	[c] 'marksl' is set
	[d] previous line begins with underlined (i.e., LR) space (or
	    maybe any other underlined character)
	[e] might only occur if inserting at the beginning of the new
	    line.
	
	-----
	5.1	bug 1st recorded, but I noticed it in earlier versions.
	
+ (28) ^U in insert mode -- only erases to point where the language of
  the text changes, not necessarily the beginning of the line.
  
	5.1	bug 1st noted; may have existed in earlier versions.
	
* (29) If you type the keys too fast, they might not be interpreted
  properly.  Up until now, vi.iv has been installed with debug & trace
  code in it.  I'm hoping that installing it compiled with the -O
  option & stripped will help this.
  
	5.1	bug 1st recorded; noticed in earlier versions.
