Small splay-tree addition

[Jeffrey A Law]

  In message < 199904030200.SAA06057@adsl-206-170-148-33.dsl.pacbell.net >you wri
te:
  > Interesting.  So, when you say `&f' for some function `f' which value
  > you get can be different depending on something or other?  Nasty.
Yup.  Actually, I've never seen it with &f, only with calls into the dynamic
loader to get the address of external routines in a shared library.  In fact,
if you make back to back calls to get the address of a routine, you get two
different values.

I haven't tried hard to see if I could get &f to give back two different 
answers, but I bet it would if I spent some time working on it.


  > But, in the case I'm using it, the comparison is only for
  > heap-allocated data (namely, the GCC `tree' type), so this isn't (yet)
  > scaring me.  None-the-less, I'll certainly add a comment, if you can
  > give a little more detail about the situations in which this won't
  > work.
OK.  This sounds safe for HPs :-)

jeff