A few random pot-shots responding to multiple posts in the thread:

 * NetBSD is not specifically for old machines.  It just happens to be
   able to run on old(er) machines because some of the developers keep
   some of the fat off.  (What amounts to "fat" depends on your point
   of view, of course.  E.g., the i386 port has apparently gained quite
   a bit from adding more drivers over the years.  Drivers for hardware
   that you don't have look like fat, but having the kernel just work
   on random hardware is a plus...(^&)

   Part of the reason that the fat is kept out is because NetBSD is
   highly portable, including targeting platforms for which there are
   no "new" machines available.  One big source tree for all targets
   helps keep the memory/CPU pigs out.  (^&  This may be what you meant
   by saying "oriented for old machines", but old machines are hardly
   the main point in NetBSD, as I understand it.  (^&

 * The tilde (~) does not begin any escape codes that I'm aware
   of.  I think that it is actually the *terminator* character for
   some escape sequences.  The escape sequences *begin* with a
   character called, strangely enough, ESC or Escape.  (^&  (0x1b,
   decimal 27...see ascii(7), I believe.)  At least for most VT100
   and ANSI style escape sequences, this is true.  I think that ssh
   actually does use ~ as an escape character.  And telnet uses
   ^] (<control>-<]>), in contrast to the usual ESC which is
   also known as ^[...

   And the ESC may be part of your problem.  The ESC that begins the
   escape sequence for keys like <F1>, <Del>, and <Home> can kick
   vi out of insert mode.  Then the following characters become
   command-mode keys; and if memory serves, the usual <Del> key
   representation ends with tilde (~), if it is not the DEL ASCII
   character.  (DEL often being assigned to the <BackSpace> key and
   the ASCII BS code not being assigned anywhere---except on systems
   where BS is assigned to <BackSpace>, and then DEL is often assigned
   to <Del>.  (^&  Did I mention that EMACS thinks that ASCII BS is
   the Help key?  I won't stoop to saying that "EMACS's Help is ASCII
   BS", since I'm above that.  But I could say it...you don't know
   how tempting it is, even though I might be punished for the pun I
   shed...(^&  No aspersions to EMACS's Help necessarily implied.
   ((void*)0) where prohibited.  Do not eat.)

   vi uses a heuristic handle escape sequences, I believe: If the
   ESC character is followed quickly enough by other characters,
   vi tries to treat the whole as an escape sequences rather than
   as a distinct <Esc> key-press followed by commands.  If for
   some reason the keys get separated far enough (or perchance the
   host computer's clock is misbehaving; (^&) vi may not figure out
   that the characters are part of one key press.  And, too, it
   has to know that your key mapping really is going to be generating
   these escape sequences.

   (The last sentence, I think, is probably not an issue in your
   case.  And the whole last paragraph may more properly be attributed
   not to vi but to termcap.  I really don't know those details.
   (^&)

 * Capital I, in command mode, will start Inserting at the beginning
   of a line.  Using <0> <i> also works, but if you count qualifiers
   as 0 or .5 keys, then <shift>-<i> is "cheaper" than 0 <i>.  (^&
   Whether vi's notion of mode-based editing is a virtue is a debate
   that I won't enter into, though.  What can be evil is switching back
   and forth between emacs and vi...the fingers don't always pay
   attention to the change.  (^&

 * Insofar as editors go, I think that it's worthwhile knowing ed and
   vi, whether you use them much or not.  There can be times when
   they are all you have.


(ramble)

Um, now that I'm more guilty than any of the rest of you: What does
this thread have to do with running -current?  (^&

-- 
  "I probably don't know what I'm talking about."  http://www.olib.org/~rkr/