Re: HP-PA ABI issues - framepointer handling

[dewar at gnat dot com (Robert Dewar)]

<<5. You can use something like dwarf2, then fall back to a slower, but
   more generic unwinder if dwarf2 stuff isn't available.  If you must
   unwind through qsort or 3rd party libraries, this is going to be
   unavoidable.
>>

Actually our analysis is that following the HP ABI on frame pointer
handling (which to me is a *very* legitimate part of the ABI, I am
puzzled that you think that frame pointer handling is out of scope
for an ABI) may actually improve performance in some cases. More
on this later.

4. The ABI places nasty restrictions on how you can generate code.

That is true of many ABI's for example, the horrible convention on  the
SPARC that all records of any size must be passed by address rather than
in registers, but gcc follows that rule. Sure, it is tempting to thumb
your nose at an ABI on performance grounds (many, indeed most, Ada compilers
have always done that in the past), but in the long run there are many
gains from rigorously following the ABI.

Anyway, we will continue to investigate here, and report our results and
recommendations.